HP-119 It therefore can be reized only in character strings (or an object type displaying as a character string), unless some of its limited forms (without separators or other details such as time zone) are interpreted as pure integers (without displaying decim point) -- but in the latter case HHMM and HHMMSS would be ambiguous (e.g. would the integer 1234 mean 001234 or 1234, which actuly represent 00:12:34 vs. 12:34:00 ?) If so, then none of the ideas about rearranging re numbers in HP ccs to represent dates (and/or times) rely satisfies the standard at l (especiy the date AND time combined notation), so why are we even trying to force it to? The HP conventions for representing dates and times as re numbers seem as close as could be fashioned to how they have most commonly been denoted in practice by users in both in USA and Europe, the shortest forms MM.DD (or DD.MM) corrsponding to MM/DD (or DD/MM), with YYYY *appended*, as was ways the cus; for time, the shortest form HH.MM, expandable into HH.MMssssss___ So where is there any ambiguity? Would there be any possibility of mistaking whether YYYY, if present at l, were prefixed (vs. suffixed) to MM.DD ? The only ambiguity is whether, if there is no YYYY or a suffixed YYYY, you mean MM.DD or DD.MM (flag -42 in 48/49, but coupled with fraction mark in HP12C and older, IIRC) When I post dates in text, and I would guess that no one has ever found any ambiguity in this, as it so appears as an obvious pattern to ISO. Google currently shows YYYY-MM-DD HH:MM:SS ZZZ (no T between date and time, though :) So look, if there is no currently proposed date AND time (combined) format for current HP ccs, nor any implementation of option year-only, year and month only, etc., why not give up trying to force this onto the re number formats currently used? Why not compose little programs (or a library) to just provide l possible conversions between re (HP) and string (ISO) formats, including, of course, the *combined* date/time string, which perhaps would correspond to a { list } of two res in the HP format. Oh, of course, since there are actuly many more minor variations in the ISO string formats, aording to how much precision you actuly want to show, whether or not you want the separatotime zones, etc., perhaps you need extra arguments (or flags?) to adjust that as well :) I have often had to interpret free-form text dates, entered by a mix of people raised in different continents (5/31 and 28/12 are unambiguous, while 4/6 is not :) When it comes to ISO-lowed year-only or year/month combinations, our databases do not low MM=0 or DD=0 (though some Sharp organizers may low MM/blank/YYYY in TODO lists :) This reminds me of language translation; at the 1964 World's Fair in New York, IBM was touting that language translation by computer was just around the corner -- so why has it flen so far backwards today, judging by Systransoft etc.? ==== > When the year is omitted in HP cculator dates xx.xxYYYY > (in many models), the current year is rly assumed. There is one speci case in models like 17B,18C,19B, > where if I enter the date in an *appointment* (arm) > as 1.01 (or any date earlier in the year than today), > then it will set my appointment for January 1 of *next* year! In the date *cculations* menu, however, > the current year is ways assumed if not specified. > JMP> What manu did you find that in? Is this not in the manu? > Well, it must be auatic, then I should've Read The Fine Manu before posting. Actuly, the Programmers Reference Manu, Advanced User's Reference, and Advanced User's Guide l say, under ->DATE: If YYYY is not supplied, the current specification for the year is used. I guess that when I use the ->DATE command, I most ways need to include the year; changing just the month or day of month would be rare. But having the current year assumed makes finding the day of year, days between two dates in the current year, and so on, especily easy. ==== M. Prange By gosh, it is! At least as far as I've checked, the ->DATE > command on the 48SX, 48GX, and 49G. What manu did you find > that in? It seems that HP beat Heiko to this idea by quite a > long time. I was aware of the behaviour that a missing year in an HP-format is > assumed as the actu year, and this is a good idea. As long as the year 0000 isn't lowed. > But HP was not aware that the year should come before the month... > I'm not sure whether there was an ISO standard for this at the time that HP started using its own date format. The only place I recl seeing a YYYY-MM-DD format back then was on Japanese blueprints. But most significant on the left certainly makes sense to me. > It is discribed (hidden) in the AUG HP48G series under remarks to > the ->DATE command (next page 3-74). > The trick works with DATE+ and DDAYS, too, which is not mentioned. ==== > That ISO standard, now that I have read its interpretation > (I did not purchase it from ISO, which only sells it for $$$), > It therefore can be reized only in character strings > (or an object type displaying as a character string), > unless some of its limited forms (without separators > or other details such as time zone) are interpreted > as pure integers (without displaying decim point) -- > but in the latter case HHMM and HHMMSS would be ambiguous > (e.g. would the integer 1234 mean 001234 or 1234, > which actuly represent 00:12:34 vs. 12:34:00 ?) It seems you are better than the working group members of the ISO. You should make a pettition to ANSI. > If so, then none of the ideas about rearranging re numbers > in HP ccs to represent dates (and/or times) > rely satisfies the standard at l > (especiy the date AND time combined notation), > so why are we even trying to force it to? Right, *if ... but I think it is a pure approach to use to different notations e.g. mm.ddyyyy and dd.mmyyy. > The HP conventions for representing dates and times > as re numbers seem as close as could be fashioned > to how they have most commonly been denoted in practice > by users in both in USA and Europe That is the problem, *two notations, and both are not that logic compared to yyyymmdd or something with a dot somewhere. > for time, the shortest form HH.MM, expandable into HH.MMssssss___ This is fine, at least because of its order for houminutes, seconds and decim seconds. > Google currently shows YYYY-MM-DD HH:MM:SS ZZZ > (no T between date and time, though :) Is google making standards? Ok, they do not apply it consequently... ... > Why not compose little programs (or a library) to just > provide l possible conversions between re (HP) > and string (ISO) formats, including, of course, > the *combined* date/time string, which perhaps would > correspond to a { list } of two res in the HP format. ==== By the way, here's another potenti ambiguity when delimiters - are omitted from dates (as so when integers are substituted): What does date 201112 mean? o 20-11-12 (permitted YY-MM-DD) = Nov 12, 2020 o 2011-12 (permitted YYYY-MM) = Whole month of Dec, 2011 . ==== > By the way, here's another potenti ambiguity when delimiters - > are omitted from dates (as so when integers are substituted): What does date 201112 mean? o 20-11-12 (permitted YY-MM-DD) = Nov 12, 2020 o 2011-12 (permitted YYYY-MM) = Whole month of Dec, 2011 > ...and that's another example why for serious date computing, the ISO date format is used. (don't have the word for 'Lohnabrechnungsprogramm' handy), and statistic test & evuation programs as parts of our production information systems, using ORACLE as database, which supports ISO standards. It doesn't er which data 'view' the user prefers, main thing is to canize date inputs to ISO, so later processing will be possible in any case. Maybe someone prefers to use a date format like YMYMYDYD, which may be a nice and cheap cryping scheme, but for practic purposes it's simply nuts ==== > By the way, here's another potenti ambiguity when delimiters - > are omitted from dates (as so when integers are substituted): What does date 201112 mean? This means you need to read section 4.6 of ISO8601 again One should not use truncated dates (there are exeptions). PS: I like Raymonds artistic date- format like YMYMYDYD. ==== Mentioned on the Museum Forum: http://www.martelinstruments.com/cased.html#MCP8800 If anyone tries one of these, then please do let us know what you think about it. ==== I recently bought an used 49g, and sometimes strange happens: it hangs for a little (2 second, or so) so when performing norm cculations. I tried to reload the firmware, but nothing changes. Are you aware of such problems? ==== > I recently bought an used 49g, and sometimes strange happens: it hangs for > a little (2 second, or so) so when performing norm cculations. > I tried to reload the firmware, but nothing changes. http://www.hpcc.org/search.php?query=1.19-6 This might help. I assume you have an old (try the command) VERSION. and if it is old Try ROM 1.19-6 or wait for 1.19-7? ==== > it hangs for a little (2 second, or so) Welcome to Garbage Collection (), which is a periodic (and unavoidable) re-organization of temporary memory. It can be made less noticeable, even in the latest ROM version: http://groups.google.com/groups?selm=3DAA0601.31A54958%40miu.edu . ==== I'm looking for a short method to abtain the numeric form of the answer in a HP49. I mean the number of the last answer, not the NUM(ANS) function. The method suggested some thread before (copy paste) seems to me quite long. For example, sometines I need SQRT(ANS), then I reize that I have to transform some sistem of unit and I have to manipulate the expression, but ANS, now, is another number, so it will be better having SQRT(numeric form): I can manipulate the expression in the history. The problem is that once pressed SQRT, copy and paste doesn't work as the cursor is fixed on the command line. Isn't there a flag or something else to obtain the number from the key (ANS)? |HIST| > I'm looking for a short method to abtain the numeric form of the > answer in a HP49. I mean the number of the last answer, not the NUM(ANS) function. The method suggested some thread before (copy paste) seems to me quite > long. For example, sometines I need SQRT(ANS), then I reize that I have to > transform some sistem of unit and I have to manipulate the expression, > but ANS, now, is another number, so it will be better having > SQRT(numeric form): I can manipulate the expression in the history. The problem is that once pressed SQRT, copy and paste doesn't work as > the cursor is fixed on the command line. Isn't there a flag or something else to obtain the number from the key > (ANS)? ==== >|HIST| >> I'm looking for a short method to abtain the numeric form of the >> answer in a HP49. >> I mean the number of the last answer, not the NUM(ANS) function. >> The method suggested some thread before (copy paste) seems to me quite >> long. >> For example, sometines I need SQRT(ANS), then I reize that I have to >> transform some sistem of unit and I have to manipulate the expression, >> but ANS, now, is another number, so it will be better having >> SQRT(numeric form): I can manipulate the expression in the history. >> The problem is that once pressed SQRT, copy and paste doesn't work as >> the cursor is fixed on the command line. >> Isn't there a flag or something else to obtain the number from the key >> (ANS)? > l I want in a future HP48 is more memory, more speed, and perhaps > seperate buttons for arrows. That's l. Coming soon? It's ready here: HP 49G > Would be cool to do some fast Saturn-compatible CPU on an FPGA... from > the energy point of view it might not be the best choice (rely not), > but it would probably easy to make it MUCH faster. yes, Dan! AND that thing would eat batteries for breakfast, lunch, dinner and then some more for snacks in between. Otherwise I welcome this idea for speeeeeed. It's about time that HP takes the speed crown again. Can it cost more than 300? ? Maybe a cheaper, smler model for poor students? If HP will not make - we will do it together!!! ==== Some remarks: - it looks like the drawn Z axis is plotted upside down - negative vues are not supported , here's a somewhat unusu announcement. DRAW3DMATRIX for the HP-48 ! http://raymond.del.tondo.bei.t-online.de/files/d3dm.zip Most 49G users (should) know what this tool is for. For HP-48 (S & G series) users: > This program plots a 3D representation of a matrix, > which can be rotated and zoomed in RE TIME. For more information, please download the archive, > and read the documentation file. Not much to add: Check it out ! ==== I did not compare it with the HP49 (until now), I just noticed the weird behaviour. However, after comparing it with the DRAW3DMATRIX command on EMU48 with 1.19-5 ROM of HP49, I conclude that the HP49 version is bugged (or is it a feature?), hence the port is too. The positive direction of the Z axis of the little axes system is plotted upside down (at least, it looks that way to me). The other reported misbehaviour was my fault, I did not adjust the Z boundaries for the plot. > - it looks like the drawn Z axis is plotted upside down > - negative vues are not supported Caspar Del Tondo schreef in bericht > Hi l, > here's a somewhat unusu announcement. > DRAW3DMATRIX for the HP-48 ! > http://raymond.del.tondo.bei.t-online.de/files/d3dm.zip > Most 49G users (should) know what this tool is for. > For HP-48 (S & G series) users: > This program plots a 3D representation of a matrix, > which can be rotated and zoomed in RE TIME. > For more information, please download the archive, > and read the documentation file. > Not much to add: Check it out ! ==== > Is there a way in sys rpl (hp-48gx) to disable the ATTN flag > so that my program can run uninterrupted? The ATTN flag is more like a student raising her hand in class; only when the teacher acknowledges and cls on her is there an interruption of what was going on, which lows the teacher to choose only a suitable and convenient moment to pause to honor the request. Thus ATTN doesn't actuly interrupt anything; a SysRPL or ML program recognizes ATTN only when it wants to, so that critic parts of system programs are never truly interrupted, except via ON+C or a CPU reset, which then forces a system warmstart for safety's sake. However, certain common UserRPL commands and program structures frequently keep testing ATTN and aborting if set, via CK0ATTNABORT or ?ATTNQUIT, which is what creates the illusion that the ON/CANCEL key interrupts the program. If you mean, can you prevent your UserRPL programs from testing for ATTN, in r you can not, but you can do something else, which is to use the IFERR structure to trap a CANCEL and then continue doing something past that point (or resume some loop structure), e.g.: << DO IFERR CLLCD I'm thinking... RAND 6. * CEIL DISP 1. WAIT THEN CLLCD Don't bother me! 7. DISP .3 WAIT END UNTIL 0. END >> You can use Wickes' IFERR Library on Goodies Disk #3 to make it much harder to break out from the above loop. If you mean, can you write a SysRPL program which ignores CANCEL, then rly l you have to do is not deliberately test for ATTN, nor use any other system entry points which do. On the other hand, if you publish time-consuming programs which neglect to bother checking for ATTN at l, other users rly start complaining about being held prisoner against their will :) . ==== > If you mean, can you prevent your UserRPL programs from > testing for ATTN, in r you can not, but you can > do something else, which is to use the IFERR structure > to trap a CANCEL and then continue doing something past that point > (or resume some loop structure), e.g.: > << DO IFERR CLLCD I'm thinking... RAND 6. * CEIL DISP 1. WAIT > THEN CLLCD Don't bother me! 7. DISP .3 WAIT END UNTIL 0. END >> One should perhaps mention that this does not eliminate a l l CANCEL interruptions because the delimiters << and >> themselves do nothing else than CK0ATTNABORT, i.e. they wait on CANCEL key press and then simply abort the program, clearing by the way the key buffer and the ATTN stack. One of the reasons to create Extended UsrRPL was to reduce the number of these mostly dispensable interrupters to give the norm user more freedom in handling interruption. > On the other hand, if you publish time-consuming programs > which neglect to bother checking for ATTN at l, > other users rly start complaining > about being held prisoner against their will :) Right! I would say such program creators are unpolite. Busby and I, we fought in vain against elimintation of an interrupt handler in the HP49 of the DownArrow key (essentily the ViewLevel1 command outside edit mode). Interrupting the DownArrow key with CANCEL was a builtin feature on the HP48. - ==== > If you mean, can you write a SysRPL program which ignores > CANCEL, then rly l you have to do is not deliberately > test for ATTN, nor use any other system entry points which do. IIRC the self-evuation code of glob names checks for the ATTN flag, so be careful to do ' ID something @ (of course you can so use another command of the @ family) instead of just ID something ==== > IIRC the self-evuation code of glob names checks for the ATTN > flag, so be careful to do If ID Name is directly evuated in a program with not own interrupt handler, CANCEL interrupts at ID Name, even if a program in ID Name does not want to be interrupted and has no interrupt handler. I never understood this behaviour. so something like 'EvNoCK:_ ID Name wouldn't help, though this were a very good place to drop any forced interruption. As to my experience, interruption can only be avoided by writing something like the lengthy :: ' ID Name @ IT EV ; instead of simply ID Name inside a program. ==== Could someone please translate the Wickes IFERR library to the HP 49G? Any other tools? ? > IIRC the self-evuation code of glob names checks for the ATTN > flag, so be careful to do If ID Name is directly evuated in a program with not own interrupt > handler, CANCEL interrupts at ID Name, even if a program in ID Name does > not want to be interrupted and has no interrupt handler. I never > understood this behaviour. so something like 'EvNoCK:_ ID Name > wouldn't help, though this were a very good place to drop any forced > interruption. As to my experience, interruption can only be avoided by > writing something like the lengthy > :: ' ID Name @ IT EV ; > instead of simply ID Name inside a program. - ==== However, it still pops out of the routines, but less easily. My intention is not to lock up the machine, but rather to provide a graceful exit. > > If you mean, can you write a SysRPL program which ignores > CANCEL, then rly l you have to do is not deliberately > test for ATTN, nor use any other system entry points which do. IIRC the self-evuation code of glob names checks for the ATTN > flag, so be careful to do ' ID something @ (of course you can so use another command of the @ family) instead > of just ID something ==== > Right! I would say such program creators are unpolite. Busby > and I, we fought in vain against elimintation of an interrupt handler in > the HP49 of the DownArrow key (essentily the ViewLevel1 command > outside edit mode). Interrupting the DownArrow key with CANCEL was a > builtin feature on the HP48. That's the first time I've heard of it... If you are referring to the decompiler not being able to be interrupted there's a good reason for it. The new ->STR is written 100% in machine language and is at least 100 times faster than the origin HP48 decompiler. So in 99% of the cases the decompilation would be finished on the HP49 before you actuly press the CANCEL key ==== Seen: > CANCEL interrupts at ID Name, even if a program in ID Name > does not want to be interrupted and has no interrupt handler. > I never understood this behaviour. Try << 'X' 'Y' DUP2 STO SWAP STO X >> Understand now? How about something like FAST, adjusted instead to disable interrupts entirely, until the program it executes completes? Of course, if the program does not complete, then you'll need a paper clip. . ==== >> Right! I would say such program creators are unpolite. Busby >> and I, we fought in vain against elimintation of an interrupt handler in >> the HP49 of the DownArrow key (essentily the ViewLevel1 command >> outside edit mode). Interrupting the DownArrow key with CANCEL was a >> builtin feature on the HP48. JA> That's the first time I've heard of it... refers to something like this: http://groups.google.com/groups?hl=en&lr=&ie=UTF-8&oe=UTF-8&threadm=3B0CFA62 .6F1CD648%40math.fu-berlin.de&rnum=5&prev=/groups%3Fas_q%3DViewLevel1%26safe % 3Dimages%26ie%3DUTF-8%26oe%3DUTF-8%26as_ugroup%3Dcomp.sys.hp48%26as_uauthors % 3Drautenberg%26lr%3D%26num%3D100%26hl%3Den JA> If you are referring to the decompiler not being able to be interrupted JA> there's a good reason for it. The new ->STR is written 100% in machine JA> language and is at least 100 times faster than the origin HP48 decompiler. JA> So in 99% of the cases the decompilation would be finished on the HP49 JA> before you actuly press the CANCEL key But, if you aidently use it on a large System RPL program, you may be waiting forever, with only ON+C to help. - ==== Hi. To late for the contest, and I haven't read the rules....but here's a smler userrpl '49 version: 69 bytes, and hopelessly slow and stupid << RCLF DATE ROT DDAYS -42 SF DATE SWAP DATE+ WHILE DUP 1 DATE+ DUP2 < REPEAT NIP END DROP IP SWAP STOF >> Chr Meland Rautenberg skrev i melding As observed by , most solutions of the chlenge presented here do > no work for dates from December 9999. Since it wasn't clearly said that > these have necessarily to be included, we agree to ignore them. First place in UsrRPL: > Dominik and Otto Praxl > with their 70.5 bytes program > << > RCLF DATE ROT DDAYS -42 SF DATE SWAP DDATE+ > FP 1 + DUP 31 DATE+ FP 1 + DDAYS SWAP STOF > First place in SysRPL: > Heiko Arnemann and Werner Huysegoms with their 50 bytes program > :: > %80 %1/ %1+ DUPROT DDAYS 42 SetSysFlag DATE+DAYS %FP %1+ %4 > %10* DATE+DAYS %FP %1+ %-1 DATE+DAYS %IP > ; > @ Their will be no offici second and third place. Werner should have > observed that in his UsrRPL program -42 SF is smler and probably > faster than 6. -7. * (6.5 bytes versus 7.5 bytes). Running programs in > approx. mode containing zints for CF, SF and other UsrRPL commands which > need res as arguments on the HP48 does not cause any problem on the > HP49. Clearly, the above SysRPL program can essentily be shortened by using > some unsupported but stable 49-pointers. We're lucky. I just got a > areas 25EC...B3C7 and 25565...4000 commented on the competition in detail which took him a lot of time - ==== Haven't seen this name in a long time.. Have I not been paying attention or has it rely been *years* since I've seen Christian Meland post to c.s.hp48?? > Hi. To late for the contest, and I haven't read the rules....but here's a > smler userrpl '49 version: 69 bytes, and hopelessly slow and stupid << RCLF DATE ROT DDAYS -42 SF DATE SWAP DATE+ > WHILE > DUP 1 DATE+ DUP2 < > REPEAT NIP > END > DROP IP SWAP STOF ==== You're right, some yeayes Nice to be recogniced Chr Aaron Wlace skrev i melding > Haven't seen this name in a long time.. Have I not been paying attention > or has it rely been *years* since I've seen Christian Meland post to > c.s.hp48?? To late for the contest, and I haven't read the rules....but here's a > smler userrpl '49 version: > 69 bytes, and hopelessly slow and stupid > << RCLF DATE ROT DDAYS -42 SF DATE SWAP DATE+ > WHILE > DUP 1 DATE+ DUP2 < > REPEAT NIP > END > DROP IP SWAP STOF > > Chr Meland ==== > 69 bytes, and hopelessly slow and stupid Christian, thank you for your contribution. At least the NIP-command is impressive. I have never used it and I think the sysRPL pendant should be SWAPDROP. With this counting gorthm the sysRPL version should be below 55 bytes (which is the requirement for the shortes sysRPL propos including flag management). > << RCLF DATE ROT DDAYS -42 SF DATE SWAP DATE+ > WHILE > DUP 1 DATE+ DUP2 < > REPEAT NIP > END > DROP IP SWAP STOF >> ==== Note that a 48 version would be not only 2,5 bytes longer (NIP= SWAP DROP), but 7,5, because WHILE-clasuses with more than one object between REPEAT and END is compiled so that the commands between are included in one object (secondary). That require 2*2,5 extra bytes. Chr Heiko Arnemann skrev i melding > 69 bytes, and hopelessly slow and stupid Christian, > thank you for your contribution. > At least the NIP-command is impressive. I have never used > it and I think the sysRPL pendant should be SWAPDROP. With this counting gorthm the sysRPL version should > be below 55 bytes (which is the requirement for the > shortes sysRPL propos including flag management). Congratulation. . > << RCLF DATE ROT DDAYS -42 SF DATE SWAP DATE+ > WHILE > DUP 1 DATE+ DUP2 < > REPEAT NIP > END > DROP IP SWAP STOF ==== > Note that a 48 version would be not only 2,5 bytes longer (NIP= SWAP DROP), > but 7,5, because WHILE-clasuses with more than one object between REPEAT and > END is compiled so that the commands between are included in one object > (secondary). That require 2*2,5 extra bytes. > Glad to see you back ! How are you doing? It's been such a long time ==== (I answer in public, to the rest of my old friends here I'm doing fine, I'm working as a businessanyst for a creditcard-company, and I am ing that. I am not using the 48/49 that much anymore, bot I miss the mini-chlenges, and jumped on this one Still working for HP? > Christian > Note that a 48 version would be not only 2,5 bytes longer (NIP= SWAP DROP), > but 7,5, because WHILE-clasuses with more than one object between REPEAT and > END is compiled so that the commands between are included in one object > (secondary). That require 2*2,5 extra bytes. > Glad to see you back ! How are you doing? It's been such a long time > ==== The complex numbers marathon has been updated and uploaded to hpcc. Storey was so kind to read it and correct l grammars. I included so l programs this time, since many people told me that it would be much better to have the whole package in one file, instead of typing on the convenient keyboard of the HP49G ==== Hi, I have been downloaded a .LIB file to my stack, the size is 24K bytes and when I try to copy from the stack and into the library I get the error message insufficient memory, probably because I need 24 + 24 K aesible at the moment.( 48G has about 32K available) So in order to decrease the need of memory I wonder if there are any way to move this file into the library instead of copying it..? Any suggestion are welcome here! ==== since you downloaded that library, you may have a variable under which te library is stored, and the stack display shows a pointer to that library. To make sufficient room: Leave the 'copy' on the stack, but delete the variable. After deletion has taken place, press 0 STO to store the library in port 0. I'm not sure if the deletion above works with that sml amount of memory left, maybe you'll have to disable LAST args/stack first. Please refer to your HP-48 manu for more details on LAST arguments. Afterwards, don't forget to enable LAST args/stack again... Finly, turn the cc off and on again. There you are Raymond > Hi, I have been downloaded a .LIB file to my stack, the size is 24K > bytes and when I try to copy from the stack and into the library I get > the error message insufficient memory, probably because I need 24 + > 24 K aesible at the moment.( 48G has about 32K available) So in > order to decrease the need of memory I wonder if there are any way to > move this file into the library instead of copying it..? Any suggestion are welcome here! ==== > Maybe JHM finds out how to do this l in UsrRPL With FLASHEV it's a built-in UserRPL command. Maybe WR finds out how to encourage people to discover how many different approaches there may be which l can aomplish a task, instead of knocking any other way but his. == > What does the HP 12C Platinum give? Today when I advanced past 253 lines of program I suddenly found that l existing GTO gave Error 4, which rendered the first 253 lines pretty useless - but indaunted I continued on and any GTO to a line over 255 worked!!!!!! This still leaves the first 253 lines useless - but as I prefer then to the 147 left I shl just clear the memory and not go over 253 lines - apart from that the unit seems pretty good - as in the first 253 lines I have done an i-solver that is more aurate than the built-in version and is faster (not hard) I did get a production unit, but I think it must be just be a lively beta unit that escaped early :) ====Seems like a byte=maximum line address -bug to me. Are you sure about the 253? could it be 254 or 255? I'm rely hoping to get a production unit. > What does the HP 12C Platinum give? Today when I advanced past 253 lines of program I suddenly > found that l existing GTO gave Error 4, which rendered the > first 253 lines pretty useless - but indaunted I continued on > and any GTO to a line over 255 worked!!!!!! This still leaves the first 253 lines useless - but as I > prefer then to the 147 left I shl just clear the memory and > not go over 253 lines - apart from that the unit seems pretty > good - as in the first 253 lines I have done an i-solver that > is more aurate than the built-in version and is faster (not > hard) I did get a production unit, but I think it must be just be a == > Seems like a byte=maximum line address -bug to me. > Are you sure about the 253? could it be 254 or 255? Program lines are made available in bunches of 7 at a time. My i-solver is 250 lines long. That leaves lines 251 and 252 and 253 with GTO 000 in. The first instruction in the new program starts at line 252, then I type the second at line 253, and if I type the third then a new bank of 7 lines from 254-260 become available. Once this happens then l GTO in the first program don't work. I changed a GTO in the first program and found that l GTO up to GTO 255 didn't work (except GTO 000 still works I think) but GTO 256 and over does. This new bank of lines cannot be deleted - the only way to fix things is clear the program and re-type, only go up to line 253, and buy another 12C for the second program Seems like a byte=maximum line address -bug to me. > Are you sure about the 253? could it be 254 or 255? Program lines are made available in bunches of 7 at a time. > My i-solver is 250 lines long. That leaves lines 251 and 252 > and 253 with GTO 000 in. The first instruction in the new > program starts at line 252, then I type the second at line > 253, and if I type the third then a new bank of 7 lines from > 254-260 become available. Once this happens then l GTO in > the first program don't work. I changed a GTO in the first program and found that l GTO up > to GTO 255 didn't work (except GTO 000 still works I think) > but GTO 256 and over does. This new bank of lines cannot be > deleted - the only way to fix things is clear the program > and re-type, only go up to line 253, and buy another 12C for > the second program I'm rely hoping to get a production unit. I'll be very interested to see what the S/N is and if anything > has been fixed. > I can confirm this happens on my beta unit. I so posted this bug report on > the museum forum. Yup mine is a production unit, received 3 weeks ago from www.educc.net in Singapore. Last week 10 units arrived in NZ. I haven't seen one but it is unlikely anything has been fixed at this stage. But I understand units gave not appeared in stores in the USA yet. When they finly do they might land like lead bloons..- however I still like it - it does show re promise. Hopefully it will be fixed at some stage. ==== , A nice bug indeed. Here it is: When program memory consists of at least 260 lines, the instructions GTO n-255 through GTO 255 cause an Error 4, n being the last located line: 260, 267 ... 393, 400 (NB: not 399). Don't you love ugly bugs? :) Not sure if there's time to include it in my review, though. BTW, anyone My pleasure instructions GTO n-255 through GTO 255 cause an Error 4, > n being the last located line: 260, 267 ... 393, 400 > (NB: not 399). ==== IS THERE AN EMULATOR FOR THE HP 48GX THAT RUNS IN THE POCKET PC HP JORNADA 540? WHERE CAN I GET IT? ==== Take a look at www.hpcc.org But try to not shout in the future, if you want to get responses... Raymond > IS THERE AN EMULATOR FOR THE HP 48GX THAT RUNS IN THE POCKET PC HP > JORNADA 540? WHERE CAN I GET IT? ==== http://persones.ya.com/leobueno/software_de_leo.htm on my iPAQ without any problems, don't know about jornada, but being both PocketPC... Paulo Pinheiro > IS THERE AN EMULATOR FOR THE HP 48GX THAT RUNS IN THE POCKET PC HP > JORNADA 540? WHERE CAN I GET IT? ==== > Does anyone know if this is actuly a problem with the solver > or is it just the gebraic expression evuator at fault? It would be desirable to isolate the cause into a simpler example for a bug report, but my feeble attempts at so doing have failed, and l I can offer is to download FIN49A from http://www.hpcc.org/search.php?query=fin49a+OR+fin421 Then do this example: - BOND CCULATIONS Step-by-step list of keys pressed: GO-> 2 CPF TYPE? U.S. Treasury (A/A) 0 XDAYS 4.281982 SDATE 6.041996 MDAT 6.75 CPN% 100 CL 8.25 YLD% left shift and then PRICE [correct answer is 87.622] When you solve for a bond PRICE, this program is run (from within the cus menu) in the HP48G version; you may independently save it to a new variable and try it now: << DSUB 'EQ' 'PRICE$' 100. ROOT DROP DSUB1 BDISP >> This works in HP48G but gets Error: Constant? in 49G. Just change 'EQ' to RCEQ, however: << DSUB RCEQ 'PRICE$' 100. ROOT DROP DSUB1 BDISP >> Now it works fine in 49G (or 48G). Even shorter: << 'EQ' 'PRICE$' 100. ROOT >> Error: Constant? in 49G [1.19-6] << RCEQ 'PRICE$' 100. ROOT >> 87.6217994005 RCEQ: << IF h 0. == THEN CL$ COUP% PF / + 1. q YIELD% * 100. / PF / + / NEG ELSE YIELD% PF / cum q - SPFV PF 'PYR' STO ncpn 'N' STO YIELD% 'I%YR' STO COUP% PF / 'PMT' STO CL$ 'FV' STO 'PV' TVMROOT * END PRICE$ + pf COUP% * PF / cum * + >> RCLF: { #7880010200605FF3h #0 #7555610A216020h #0 } There are sever other places in the Bond and Stock subdirectories where 'EQ' (or another variable name) had appeared as an argument to ROOT, and l failed in 49G [1.19-6], unless changed to put the actu expression (in this case a program) on the stack, in place of the name of the variable containing it (in the 48G it didn't er which way this was done). l of the origin 48G programs and menus may be downloaded from FIN421B as found from the above URL. ==== Is there ANYWHERE where I can order an HP 82216a 1 MB expansion card for the HP48GX cculator? Are there any blank 4Mb cards? I did find one in Germany at http://uuhome.de/oklotz/order_e.html but I'd like to know if there are others. TIA. Order the klotz cards. There are sever people in this newsgroup who swear by them. There is an error in the hardware design which prevents direct use of a 4Meg card unless you perform a bank swap of the last 128k bank, as klotz does. For this reason, there are no HP cards that size, and few third party cards. Klotz is the only 4M I have seen. For vue and price, you would be better off with two 2M cards and manuly swap them as needed. Note that 2M is extremely hard to fill. Do not put a memory card larger than 128K in slot 1!!!. ==== Try www.cynox.de. This is where I got mine (1MB) from. Rely cheap too. ==== Is this the HP 82216A or a 3rd party card? If you got a 3rd party card, do you know if it works EXACTLY as the 82216a would? I ask because the 4MB card that I found seems to offer 32 banks of 128K bytes, which tells me that it's like having 32 128Kb cards instead of 1 one ==== GXC1024. > Is this the HP 82216A or a 3rd party card? If you got a 3rd party card, do > you know if it works EXACTLY as the 82216a would? I ask because the 4MB card > that I found seems to offer 32 banks of 128K > bytes, which tells me that it's like having 32 128Kb cards instead of 1 one > you can give me. The 1MB card is like having 8 128KB cards unless you tk about the HP 49G FlashRAM which complete hides this fact from the user. PS: without a hardware switch there might be trouble aessing the last ports in that 4MB card. http://uuhome.de/oklotz/4mb_d.html (28 statische Ports + 2 umschtbare Port-Paare) (28 static Ports + 2 switchable Port-pares) ==== Well, I've never owned a 1Mb card for this cculator before, but my impression is that if you put the HP 82216a into it, you get port added to your cculator, and this port can hold 1 Mb of memory. And you can merge the 1Mb memory with the cculator as you can the 128Kb memory with your cculator without bankswitching. Is that right? Does anyone know if the card at http://www.cynox.de/eng/index.html?cc is the hp 82216a or if it behaves EXACTLY like the hp 82216a card? > GXC1024. > Is this the HP 82216A or a 3rd party card? If you got a 3rd party card, do > you know if it works EXACTLY as the 82216a would? I ask because the 4MB > card > that I found seems to offer 32 banks of 128K > bytes, which tells me that it's like having 32 128Kb cards instead of 1 > one > you can give me. The 1MB card is like having 8 128KB cards > unless you tk about the HP 49G FlashRAM > which complete hides this fact from the user. PS: without a hardware switch there might be trouble > aessing the last ports in that 4MB card. > http://uuhome.de/oklotz/4mb_d.html > (28 statische Ports + 2 umschtbare Port-Paare) > (28 static Ports + 2 switchable Port-pares) ==== Seen: > you can merge the 1Mb memory with the cculator > as you can the 128Kb memory with your cculator No, not for user variables and Port 0; the 48GX MERGE[1] command works only for 32K or 128K in slot 1. > [is Cynox card exactly like HP card?] Epson created the card design, as memory for electronic typewriteand HP later utilized the same cards as cculator add-on memory. l that the card is required to do is to respond to its address lines and act like a contiguous blob of memory, but the cc can not actuly use that large a blob, because each of the 20-bit addresses used by its CPU references a single nibble (hf-byte), and the intern cc hardware architecture therefore interposes a memory controller, and has decided to low no more than a 128_Kbyte region of any card to oupy 1/4 of the CPU's virtu address space at any one time. So l of the restrictions on what you can do with the card are imposed by cculator design, while most cards remain plain simple memory blobs, l ike. One company, however, equipped some cards with an on-board controller and manu switches to make just one 128K section visible at a time to the card connector's fixed address lines; this card was for the 48SX, which aepted only 128K max. cards, and therefore it packaged a bunch of virtu separate cards into one physic card, with the user manuly selecting just one 128K bank at a time. 48GX cards, by contrast, have the bank selection auated by the cc's memory controller, which does it l via intern programming. Cynox so makes its 4_Mb card with a manuly switchable mapping of the very highest banks, basicly because the 48GX made a mistake in counting memory banks, and can't use the very last 128K out of 4_Mb. HP itself never ordered such large cards from Epson, its OEM, so it never had to de with that last bank problem -- a greater problem, in fact, is the very large time that it may take to inventory a 4_Mb card every time the cc switches on, which is done as a safety precaution against a user swapping cards under the cc's nose while it is off, and then invoking library functions whose locations on the card are saved in static tables which would become invid if the card was changed without telling the cc to rebuild its static tables. So you can buy Cynox cards; in fact, they must know more about making cards which work in 48GX than anyone left at HP or Epson :) The only things you may not get in non-Epson RAM cards are some shielding (covers) and an Epson-patented shutter to cover the connector edge of the card; the quity (or even presence) of write-protect and other switches might so vary among l who ever manufactured cards, and the type and longevity of on-board card batteries, and the connector specs and quity might so vary between brands. One more little thing might differ Epson designed its RAM cards to sort of short-circuit their inputs when extern voltages are sensed to be low (anti-static protection); though HP therefore ordered slightly modified cards from Epson for its ccs, origin Epson cards meant for typewriters (and Grid computers) may short out an HP cc if by bad luck the cc batteries and on-board power supply dip below the point at which the card is set to clamp its inputs, so genuine Epson cards not made for HP might be dangerous to use. But note that Cynox is not Epson, so no problem there at l ;) . ==== > Well, I've never owned a 1Mb card for this cculator before, but my > impression is that if you put the HP 82216a into it, you get port added to > your cculator, and this port can hold 1 Mb of memory. And you can merge > the 1Mb memory with the cculator as you can the 128Kb memory > with your cculator without bankswitching. Is that right? Does anyone know > if the card at http://www.cynox.de/eng/index.html?cc is the hp 82216a or > if it behaves EXACTLY like the hp 82216a card? * No! Main RAM is 128KB including intern port 0 Slot 1 is port 1, 128KB RAM, which can be merged to get the maximum 256KB Main RAM Slot 2, holding a 1MB RAM card is port 2,3,4,5,6,7,8,9, each 128KB which can not be merged into the main RAM * Yes! You can use any Cynox/Klotz/etc RAM card designed for the 48GX PS: Please, read chapter 28 from your User's Guide ==== > But I'm not sure that extending the Gregorian cendar back > before 1582-10-15 would rely be useful for anything. > [major snips for bandwidth conservation] Those who do ogic research for themselves or others would > certainly find good 'dating' prior to 1582-10-15 useful. I agree, but I think that when you have records dating from before 1582-10-15, you can be quite sure that the Gregorian cendar wasn't used for them, unless perhaps someone converted them to a proleptic Gregorian cendar when transcribing them at some later date. > Especily if > you thought a remote forebear held property. I suppose that we probably l have both royty (especily if Charlemagne was as prolific as I've read) and slaves, and both saints and scoundrels, in our ancestry. Collecting inheritances left by such remote forbears may be a bit of a problem. Let's ignore any debts that they may have left unpaid, as well as any punishments that they may have avoided. ==== > For the same reason we have HH.MMSS and then decim hours > => we should have YYYY.MMDD and decim years > as ->H and ->HMS, we should have ->YR and ->YMD > Do you mean that it should convert 2003.31415926 > to year, month, day... and maybe any fraction left over > as hhmmss... ??? > No! > I was not clear enough! > I so got yet another idea so here we go: > INPUT FUNCTION OUTPUT > 2003.0530 YMD2YD 2003.15 @ YYYY.DDD (days so far) > 2003.0530 YMD2YR 2003.4109589 @ YYYY.yyy...(decims) > 2003.150 YD2YMD 2003.053 @ YYYY.MMDD (date) > 2003.150 YD2YR 2003.4109589 > 2003.41095890411 YR2YMD 2003.053 > 2003.41095890411 YR2YD 2003.15 The decim years can be used for plotting dates vs. vues At first, I worried that you were contemplating applying trig functions to dates. Time for Veli-Pekka to take a nice long rest in a padded room? { It might be best programmed by finding the day of the year, the number of days in the year, dividing them, and adding the result to the year. But that leaves the problem that a fraction of a leap year is not the same length of time as the same fraction of a norm year. Perhaps better to use the number of days (or seconds or minutes, if time information is so available) from a particular time, for plotting purposes? I could now add the functions to change to/from YearWeekWeekday format YMD2YWD YWD2YMD YD2YWD YWD2YD YR2YWD YWD2YR OS could so use units for these: 12.30_hms 1_h CONVERT => 12.5_h 2003.0530_ymd 1_yddd CONVERT => 2003.150_yddd or YMD2YDDD instead of the CONVERT (because of yards as 1_yd I had to change the 1_yd to 1_yddd) (so so that command names reflect the units: YDDD instead of YD) > For the same reason we have HH.MMSS and then decim hours > => we should have YYYY.MMDD and decim years > as ->H and ->HMS, we should have ->YR and ->YMD > Do you mean that it should convert 2003.31415926 > to year, month, day... and maybe any fraction left over > as hhmmss... ??? > No! > I was not clear enough! > I so got yet another idea so here we go: > INPUT FUNCTION OUTPUT > 2003.0530 YMD2YD 2003.15 @ YYYY.DDD (days so far) > 2003.0530 YMD2YR 2003.4109589 @ YYYY.yyy...(decims) > 2003.150 YD2YMD 2003.053 @ YYYY.MMDD (date) > 2003.150 YD2YR 2003.4109589 > 2003.41095890411 YR2YMD 2003.053 > 2003.41095890411 YR2YD 2003.15 > The decim years can be used for plotting dates vs. vues At first, I worried that you were contemplating applying trig functions > to dates. Time for Veli-Pekka to take a nice long rest in a padded room? > { It might be best programmed by finding the day of the year, the number > of days in the year, dividing them, and adding the result to the year. > But that leaves the problem that a fraction of a leap year is not the > same length of time as the same fraction of a norm year. Perhaps better to use the number of days (or seconds or minutes, if time > information is so available) from a particular time, for plotting > purposes? ==== > That ISO standard, now that I have read its interpretation > (I did not purchase it from ISO, which only sells it for $$$), > http://www.pvv.ntnu.no/~nsaa/8601v2000.pdf After reading the ISO standard above I'm not so sure I should take > a nice long rest in a padded room unless if that's where the ISO group is. I wouldn't be surprised to learn that any standards committee delegates were ready for a nice long rest in a padded room by the time that a standard was agreed on. Notice how much room for variations ISO8601 lows; I surmise that a lot of compromises were made before a consensus was reached, and doubt that anyone was entirely happy with the result. the fin draft) in a note where they surely meant proleptic. I hope that none of them has a missile and my geographic coordinates. I remember reading a standard, I believe by ANSI on circularity (roundness). They spent years hammering out a standard only a very few pages long (less than a page per year), with a much longer appendix of unresolved issues, differing viewpoints on them, and some possibilities for actuly measuring this characteristic. I've heard tes (perhaps apocryph) of days spent in an uncomfortable room with some delegates heatedly debating the placement of a comma, whether a word should be the or a/an, or a word should be in or into, or in or within, and such seemingly minor but perhaps meaningful points when it comes to standards. Imagine the problems when it's an internation committee instead of merely a nation committee. It's amazing that they ever reach a consensus; perhaps it's a er of agreeing to flexibility through sheer exhaustion. I'd think twice (at least) before volunteering for such duties. > I could now add the functions to change to/from YearWeekWeekday format > YMD2YWD YWD2YMD > YD2YWD YWD2YD > YR2YWD YWD2YR OS could so use units for these: > 12.30_hms 1_h CONVERT => 12.5_h > 2003.0530_ymd 1_yddd CONVERT => 2003.150_yddd Ok, if: 2003.0530_ymd 1_yddd CONVERT => 2003.150_yddd then: 2004.0530_ymd 1_yddd CONVERT => 2004.151_yddd True? A difference in the fraction part? If we used a YYYY.yyyyyyyy format, where .yyyyyyyy represented a fraction year, then: 2003.0530 would convert to 2003.41095890 but: 2004.0530 would convert to 2004.41256831 Again, a difference in the fraction part. Of course for fixed length time periods such as weeks, days, hours, minutes, and seconds, these problems don't our (except possibly when a leap second or skipped second ours). But for variable length time periods such as months or yeaproblems seem to be inevitable, and if we try using average lengths for these time periods, the problems seem to get worse instead of better. Again, I'd think it much more useful to convert to and from a days from a particular day format. > or YMD2YDDD instead of the CONVERT > (because of yards as 1_yd I had to change the 1_yd to 1_yddd) > (so so that command names reflect the units: YDDD instead of YD) > For the same reason we have HH.MMSS and then decim hours > => we should have YYYY.MMDD and decim years > as ->H and ->HMS, we should have ->YR and ->YMD > Do you mean that it should convert 2003.31415926 > to year, month, day... and maybe any fraction left over > as hhmmss... ??? > No! > I was not clear enough! > I so got yet another idea so here we go: > INPUT FUNCTION OUTPUT > 2003.0530 YMD2YD 2003.15 @ YYYY.DDD (days so far) > 2003.0530 YMD2YR 2003.4109589 @ YYYY.yyy...(decims) > 2003.150 YD2YMD 2003.053 @ YYYY.MMDD (date) > 2003.150 YD2YR 2003.4109589 > 2003.41095890411 YR2YMD 2003.053 > 2003.41095890411 YR2YD 2003.15 > The decim years can be used for plotting dates vs. vues > At first, I worried that you were contemplating applying trig functions > to dates. Time for Veli-Pekka to take a nice long rest in a padded room? > { > It might be best programmed by finding the day of the year, the number > of days in the year, dividing them, and adding the result to the year. > But that leaves the problem that a fraction of a leap year is not the > same length of time as the same fraction of a norm year. > Perhaps better to use the number of days (or seconds or minutes, if time > information is so available) from a particular time, for plotting > purposes? ==== I uploaded the program for converting the HP48 graphics objects into BMP bitmap images to hpcc.org couple of minutes ago. It includes the so be downloaded from: http://www.physic.ut.ee/~/english/hpsoft/grob2bmp-1.0.tgz Best wishes, Tiismus ==== The HP 48G+ has an infrared port that can be used to connect two similar cculatoor a cculator to an HP printer. My question is: can I so connect the cculator to a laptop equiped with an infrared port? I know about the seri port option, but if the IR would so work it would be rey good for me. ==== Unfortunately the IR interface protocol used in the HP cculators is not IrDA compatible. )-: This is one more thing that HP should change in order to make the HP 39G/48G/49G cculators perfect. > The HP 48G+ has an infrared port that can be used to > connect two similar cculatoor a cculator to an HP printer. My question is: can I so connect the cculator to a laptop equiped > with an infrared port? I know about the seri port option, but if the > IR would so work it would be rey good for me. > ==== Yes, I was afraid that would be the case. I wonder if this could be circumvented by programming the cculator to behave in a different way. The problem is I'm not an expert... for answering! Unfortunately the IR interface protocol used in the HP > cculators is not IrDA compatible. )-: This is one more thing that HP should change > in order to make the HP 39G/48G/49G cculators perfect. ==== Yesterday I spoke to the swedish importer of the HP Çcculators and was told that the new name will be HP 49+ and that it will be released in mid/late august, here. Just my few pennies i .85vermark ==== > [took derivative of] tan(x) on my HP49, and then > idly pressed INTVX, thinking I'd get tan(x) back again. > Instead, it returned 'tan(x)-atan(tan(x))' Maybe it actuly returned 'tan(x)-atan(tan(x))+x' Anyway, setting flag -113 [Linear simp off] does help, which shows how *not* simplifying can rely be simpler :) I wasn't able to get the above to simplify after the fact, either (give or take an arbitrary constant), though it was very easy to complexify beyond recognition, but I am a novice CAS user, and will check back to see whether anyone posts a way to do that. . ==== > the solver application and the function plotter share the magic > variable 'EQ' but disagree on their interpretation of its contents > when they are - for example - something like f(x)=g(y) and you want > to plot an implicit variable. This disagreement can be fixed > by the user with a line of code. What's that line of code, then? In a FUNCTION type plot, the plotter convention has been that the left and right sides are to be plotted as two distinct, separate expressions (unless the left side is simply a glob name, in which case it is ignored). This convention seems to arise from the fairly common notion that you might want to look for a solution to 'f(x)=g(x)' [rather than 'f(x)=g(y)'] by seing where the two expression plots intersect. If the independent variable of a FUNCTION type plot is 'x' then g(y) does not depend on 'x' unless variable 'y' contains a formula or program which depends on 'x'; note that the plotter is rely not interested in the name of a dependent variable, except to label axes, since in a FUNCTION type plot, the vertic vue is the expression's vue, and doesn't even need to be identified with a name at l. > this could have been fixed in any firmware release > in the last fifteen years. Again, where is that line of code, and let's see whether fixing it does not break the origin capabilities, so that one particular problem is not catered to at the expense of no longer supporting the origin (and perhaps more usu) set of problems. An increase of rity which is a superset of the origin functionity would be great please post it, with a little more detail, so that its reity might be verified. > So the trade-off is: keep a stable software platform > or make (for the sake of this example) > plotting implicit variables easier for new users. Tough choice: > stable firmware or new users (grossly oversimplified, of course)? Please detail how to implement your improvement; only then can we evuate how it might affect ordinary function plots keep in mind that this plot type does not even refer to a dependent variable name at l, as it actuly varies an independent variable over a range, and plots the vue of an expression, not the vue of another variable. It seems to me that the type of plot you want is rely an entirely different type (closer to PARAMETRIC, for example). The ideas you put forth seem to be suited to an entirely new underlying architecture, in which the existing 48/49 is just partly simulated in one part (or omitted entirely); not a logic extension as a superset, with any magic fix that could have been incorporated in any firmware update in the last 15 yeaas you assert, unless, for this particular issue, you mean only that another plot type could have been added. -[]- . ==== like the HP 19B* (* = wild card) For what it's worth, the last time that I had some bank business complicated enough that I had to de with a bank officer, she had a computer running MS Windows XP. She brought up an application, but used an HP cculator, I don't know which model, except that it was a clamshell, like a 28 series. She did the cculations on the cculator and printed to an HP 82240, ripped off the printout, and typed the numbers into the fields of her computer application. How many HP48 > users actuly read chapter 29 (programming) of the '48 user guide? > I order to use a programmable cculator they should!! Quite so; if they can't be troubled to Read The Fine Manu, then they'd be better off saving some money and buying a non-programmable instead. ==== > But if Dond Knuth is correct in > suggesting that only around 2% of the population are natur > 'gorithmic thinkers' That sounds terribly interesting! I tried Google, but couldn't find any more context. Do you have a reference? I'd very much like to read something by Knuth on that topic. ==== > Could you supply some info or link to to the HP49Gii as I haven't been able > to find anything concrete about it. I'm hoping that it's more than just > another user fantasy. This discussion started by somebody writing about what he would like to see > in a new HP48. > obvious that a 5Ghz CPU didn't make much sense) No new product has been announced by HP, as it happens every once and a > while these are just unfounded rumours I've been told by an offici for HP France that HP would release 6 new models this year, starting with the 12C Platinum, the financi model. Furthermore, my correspondant refered to the future HP49G. I don't know exactly what he meant for this, but there's so a news from the offici site : http://www.hp.com/cculators/news/index.html which announces models with higher programming possbilities and greater memory. I so remember Cyrille tking about new amazing models, and you know he's still working for the cculator department of hewlett-packard. Yoann D.8esir (http://membres.lycos.fr/hp49g/english.html, http://www.hp-sources.com yoanndesir@yahoo.com) ==== > Could you supply some info or link to to the HP49Gii as I haven't been able > to find anything concrete about it. I'm hoping that it's more than just > another user fantasy. > This discussion started by somebody writing about what he would like to see > in a new HP48. however > obvious that a 5Ghz CPU didn't make much sense) > No new product has been announced by HP, as it happens every once and a Note: * Announced by HP * For the first time ever I must disagree with you There have no other announcement (except for the HP 12C Platinum) The HP web site you refer is for re, but there are no model numbers so we can't know what and when, but HP is still in business. (-: I'm dreaming of 15C Diamond with even greater speed and memory!! Currently I'm waiting for my HP with 5GHz CPU and on the contrary to which is going to sit on my desktop and have a Tejas T-slot 5MHz CPU and my EMU49 & Yorke will both fly like never before!!! (naturly I meant it as a joke, some people rely think I was tking about a battery-operated cculator) > while these are just unfounded rumours > I've been told by an offici for HP France that HP would release 6 > new models this year, starting with the 12C Platinum, the financi > model. > Furthermore, my correspondant refered to the future HP49G. > I don't know exactly what he meant for this, but there's so a news > from the offici site : > http://www.hp.com/cculators/news/index.html > which announces models with higher programming possbilities and > greater memory. > I so remember Cyrille tking about new amazing models, and you know > he's still working for the cculator department of hewlett-packard. > Yoann D.8esir > (http://membres.lycos.fr/hp49g/english.html, > http://www.hp-sources.com > yoanndesir@yahoo.com) ==== > No new product has been announced by HP, as it happens every once and a > Note: * Announced by HP * For the first time ever I must disagree with you > There have no other announcement (except for the HP 12C Platinum) > The HP web site you refer is for re, but there are no model numbers > so we can't know what and when, but HP is still in business. (-: I didn't say the 6 cculators were officily announced by HP, but I'm confident the information I got is true, because the one who told it to me works for HP. So, await new powerful cculators for september. ==== Yoann! There are no *offici* HP announcements on any new model numbers and/or dates so far (except fot the HP12C platinum) You said that you disagree on Avenard, but he clearly tks about offici HP announcements: ************************************* > No new product has been announced by HP, as it happens every once and a > while these are just unfounded rumours ************************************ You did say the above. What I now think is (unless your clarify things once more) that you disagree on the unfounded rumours part?? PS: Nina will soon use a new HP system (a 58GX?) that has RPN/G cculator OS and a CAS (similar to 49G) and is operated on batteries !!! and has a Gigahertz class CPU (WOW) (again I'm gonna fool some of you with this >:-P) > No new product has been announced by HP, as it happens every once and a > Note: * Announced by HP * > For the first time ever I must disagree with you > There have no other announcement (except for the HP 12C Platinum) > The HP web site you refer is for re, but there are no model numbers > so we can't know what and when, but HP is still in business. (-: I didn't say the 6 cculators were officily announced by HP, but > I'm confident the information I got is true, because the one who told > it to me works for HP. > So, await new powerful cculators for september. ==== > But if Dond Knuth is correct in > suggesting that only around 2% of the population are natur > 'gorithmic thinkers' That sounds terribly interesting! I tried Google, but couldn't find > any more context. Do you have a reference? I'd very much like to read > something by Knuth on that topic. Reference is gorithms in Modern Mathematics and Computer Science and originly appeared in Lecture Notes in Computer Science 122 (1981). Currently in print as Chapter 4 of: Selected Papers on Computer Science Dond E. Knuth (CSLI Lecture Notes, no. 59) ISBN 1-881526-91-7 dd. ==== > > the solver application and the function plotter share the magic > variable 'EQ' but disagree on their interpretation of its contents > when they are - for example - something like f(x)=g(y) and you want > to plot an implicit variable. This disagreement can be fixed > by the user with a line of code. What's that line of code, then? Come on ! You pointed out yourself that the user can work around this problem with a one-line invocation of the ROOT command only ten days ago. > In a FUNCTION type plot, the plotter convention has been that > the left and right sides are to be plotted as two distinct, > separate expressions (unless the left side is simply > a glob name, in which case it is ignored). This convention seems to arise from the fairly common notion > that you might want to look for a solution to > 'f(x)=g(x)' [rather than 'f(x)=g(y)'] > by seing where the two expression plots intersect. This could so be achieved by placing {f(x) g(x)} as a list of two expressions into EQ. Equly, if 'f(x)=g(x)' was interpreted by the plotter as 'f-g' - i.e. the same way as the solver application interprets it - then the zeros of the resultant single curve would so point out the solutions of 'f(x)=g(x)'. So why break the 'implicit variable' interpretation of f()=g(), in order to make it possible to plot two curves, when there is another way of achieving that two curve plot without encountering that problem? I can only assume that this decision has a 'historic' background, or that the solver and plotter code began life as very separate pieces of code. > fixing it does not break the origin capabilities, It would change the plotter's interpretation of vues of the 'EQ' variable like 'F()=G()' to match that of the solver application - so one can be pretty sure that it will break _someone's_ code. And this is the core problem of backward compatibility: no-one will seriously argue that a slightly simpler way of plotting two curves is a sound reason for breaking the obvious 'implicit variable' interpretation of f()=g(), but if it breaks backward compatibility... > The ideas you put forth seem to be suited to an entirely new > underlying architecture, in which the existing 48/49 is > just partly simulated in one part (or omitted entirely); > not a logic extension as a superset, with any magic fix It's possible to put object systems on top of existing architectures. The 48/49 looks like a pretty good candidate for this sort of extension. For example, the existing tag-based dispatch code could be extended to recognise a 'user defined object' type tag and hand off evuation to an open table-driven dispatcher. There are other possibilities. I don't think that would break much in the way of existing code, but it _would_ provide an opportunity to tailor the machine without being saddled with 2k-era design decisions. ==== > So why break the 'implicit variable' interpretation of f()=g() The word break seems to imply that you think there is some existing feature ready in the cculator to do this, but there is as yet no auatic invocation of any numeric solver during any plot, so what is there to break? > in order to make it possible to plot two curves > when there is another way of achieving that two curve plot > without encountering that problem? I don't see any problem yet, since there was never any other feature there to begin with. > I can only assume that this decision has a 'historic' > background, or that the solver and plotter code began life > as very separate pieces of code. Plotting and numeric solving of the same expressions *are* related (if the user wants to use both on the same problem) -- the plotter even has menu functions for invoking the numeric solver, to find roots and intersections. The only difference in interpretation is that since the numeric solver has to produce a single numeric vue for the entire expression, even if it is an equation, the equs sign acts like a minus sign in the numeric solver, just as it ways does with the ->NUM command; thus it represents a solution for an intersection, which seems to me to be very compatible with and complementary to the plotting application, as designed. The plotter, meanwhile, is perfectly capable of plotting multiple expressions on the same screen, and the Wickes gang seem to have felt that this would be a more commonly desirable feature for equations like f(x)=g(x) than to let the user plot only the difference 'f(x)-g(x)', rather than to see both graphs and their intersection. By the way, as said previously, 'f(x)=g(y)' would *not* represent, by itself, any implicit equation to plot, because 'y' is not any sort of default name for the vertic axis of a FUNCTION type plot (actuly, the vertic axis is better thought of here as nameless, simply associated with the entire expression being plotted). Now, you suggest that a user should be told instead to make a list of two separate expressions if he wants to be able to see the two plots. So let's say that he listens to you and makes this two-expression list; okay, now he wants to find the intersection how does he do that? Unfortunately, now neither the plotter nor the numeric root-finder can get the intersection any more, so speaking of things being broken, now they rely are! So, my friend, what I see here is another case of the origin designers being quite visionary and holistic, and folks who came in afterwards with narrowly conceived tweaks being more likely to break what had started out as a totly well-conceived and fully integrated, harmonious package, from the ground up. You can see this in various cracks l around the HP49G, I'm afraid it is very fine to have added a capable CAS, a Filer, more memory, flash ROM with user storage, and other goodies, but at the same time, it doesn't l smoothly blend with the architecture of the origin house, perhaps since no origin architect seems to have been aboard to lend a hand (and a visionary eye) in the transition. > It's possible to put object systems on top of existing architectures. Maybe so therefore the next step would be to show how, in practice, or to leave it as a suggestion to consider or perhaps, for WR to implement in his next new library ;) ==== implicit function plots: I meant to add the thought that instead of trying to force implicit functions on top of the existing FUNCTION plot type, where I have been suggesting that it doesn't fit (and that trying to force it could even break the origin application), why not cast your desire as a new plot type instead, where you are free to structure it to fit the problem, without meanwhile destructuring the origin plot type :) There are ready some ternate plotter libraries or programs; perhaps, unknown to us, this whole thing might ready have been addressed, or if not, the author(s) might want to consider incorporating additions? Or, perhaps it's rely just a speci case of PARAMETRIC plot, which ready exists, in which the parameter could be either one of the axis variables, and the other variable would still need to be numericly solved at each point (don't forget to so demand a higher speed CPU for this sort of thing :) ==== I would like to see my cculator fully integrated into a math software package for my computer. The computer software would use l of the same commands as the cculator, but with a full screen and no memory/processor limitations (besides the ones on the computer). This is only if I want. I should so have the ability to have it display an emulator just like EMU49 (or EMU58 in this case). The cculator will be able to trasfer objects much more quickly. As soon as I plug it in to the computer and click connect, the computer will have the cculator's history, and update any variables I've created or purged. In the HP tradition, it will ask me if I want it to do this before it does. Using technology similar to MIDI (Music Instrument Digit Interface), whenever I press a button on the cculator, it executes a command on the computer. Basicly my cculator will be like another keyboard. At the end of my session, I decide which objects I want to keep on the cculator. Maybe there's a little window with the directory tree that I can drag and drop things into. Of course, there's plenty of RAM, so I won't have to be too conservative about this. The software will have at least two windows: One window diaplays the stack. If I press 2 4 + it will display '2+4' unless I tell it to evuate it. If I press 0 1 X 2 ^ X INT, it will display 'INT(0,1,X^2,X)' on the stack unless I tell it otherwise. This way I won't lose the numbers I used to get my answers. I can use this to chase mistakes that I might have made when I was entering my numbers. The other window displays whatever is in the stack level that is selected (defaulted to 1) in whatever format it thinks is appropriate. For example, if I press 2 4 + it will display '6'. If I press 0 1 X 2 ^ X INT, it will display the textbook format of the integr from 0 to 1 of X^2 with respect to X. This window will so display graphs, but I would have a lot more control over how my graphs look. The cculator will still function like l other HP cculators. ==== I'm trying to integrate x^0.16, but my 49g gets stuck.... int(0,0.3,y^0.16,y) where int is the integr sign... I get no answer!! numeric or symbolic.... I get no answer!! numeric or symbolic.... Well you should get an answer, though it takes a long time on the HP49G to cculate this numericly. Just enter the integr a bit different. Make sure you are in exact mode and that 'X' (or the variable you are using) don't hold any numbers. 'int(0,3/10,X^(16/100),X)' where int ist the integr-sign.You get the exact solution first, just press [R-S]-[NUM] to get the approximate solution (I guess you know this ready). ==== To make integration time shorter you may want to use SCI 4 (or FIX 0 or ENG 9 or) The Numeric Integrator use the display auracy information as the limiting factor. STD uses l 12 decims. IERR will tell you the maximum estimated error. Usuly you get one more correct decim than acquired. You may want to test this by using simple known integrs with varying digits for the SCI .You get the exact solution first, just press > [R-S]-[NUM] to get the approximate solution (I guess you know this ready). ==== What are your flag settings? I entered it in exact mode, and it gave me a warning telling me that it was integrating on approximate mode. The answer it came up with was 'int(0.,.3,exp(.16*ln(Yt)),Yt)'. This is the 49's way of saying, I can't find a symbolic solution for this, but if you want me to, I'll find a numeric approximation for it. If I press [->NUM] it cculates it (maybe) but takes a re long time. I got tired of waiting after about 30 mins and stopped it. FYI, you usuly won't get a symbolic answer when evuating an expression with decims in it. I've found that the 49 seems to prefer intergrating in exact mode. I typed in the same integr, replacing the decims with fractions: 'int(0, 3/10, Y^(16/100), Y)' and I got an exact answer (15*XROOT(25,3/10)^4/58) in 8.3116 seconds. [->NUM] gives me .12330580085 in .3494 seconds. This is significantly better than the time it takes for the 49 to approximate the answer. When you say you don't get an answer, do you mean that the cculator ignores the [EV] command? When you press [ON], does it say Interrupted!? Does it change it to the weird looking thing with Yt in it? I think you might be stopping it before it has enough time to cculate the answer. Either way, I would recommend integrating in exact mode whenever possible.==== > I've found that the 49 seems to prefer intergrating in exact mode. I > typed in the same integr, replacing the decims with fractions: using XQ > 'int(0, 3/10, Y^(16/100), Y)' and I got an exact answer (15*XROOT(25,3/10)^4/58) in 8.3116 seconds. > [->NUM] gives me .12330580085 in .3494 seconds. This is significantly .21... not .12... > better than the time it takes for the 49 to approximate the answer. When you say you don't get an answer, do you mean that the cculator > ignores the [EV] command? When you press [ON], does it say > Interrupted!? Does it change it to the weird looking thing with Yt > in it? .90 did continue after the 'int(0.,.3,exp(.16*ln(Yt)),Yt)'. using ->NUM and waited long enough to get the same answer PS: I just started the origin integr & { ->NUM } HEAD MEM DROP then TEV from CATog...let's wait and see how long it takes... ==== > I'm trying to integrate x^0.16, but my 49g gets stuck.... > int(0,0.3,y^0.16,y) > where int is the integr sign... > I get no answer!! numeric or symbolic.... > Well you should get an answer, though it takes a long time on the HP49G to > cculate this numericly. I'd cl that a bug. It should be instantaneous. > So should be a sml numeric matrix inversion, but that's not so. Unfortunately computing does take time. After 1h 1min and 51 seconds I got the answer in STD I may put out a table for different answers/timings in different SCI formats Where is Nick when he is needed? PS: after XQ the integration took less than 9 seconds << XQ EV ->NUM >> was less than 9.5 seconds ==== [re the HP cculator site giving Page not found with Netscape] > It works fine in Mozilla, Konqueror and Opera... > Maybe Netscape is the problem? Turn off JavaScript and it *then* works fine in Netscape, too, so not likely; I suspect that the site webmaster just made a trivi mistake, re-directing to an invid URL on just this one browser/feature combination. ==== > IFTE is an operator, not a function, and cannot be used in > the built-in table operation. > That is incorrect ; IFTE is a function and can be used in the table or any > plotting > If it doesn't then there's a bug.. I believe a problem was listed in > 1.19-6 > that has been fixed in 1.19-7 It doesn't work correctly in 1.19-3, but this may not be for the reason I stated. ==== Your quota points to wrong person, Vir. I asked JYA to check if it rely works in 1.19-7 Be more careful, please... > IFTE is an operator, not a function, and cannot be used in > the built-in table operation. > That is incorrect ; IFTE is a function and can be used in the table or any > plotting > If it doesn't then there's a bug.. I believe a problem was listed in > 1.19-6 > that has been fixed in 1.19-7 It doesn't work correctly in 1.19-3, but this may not be > for the reason I stated. ==== > It doesn't work correctly in 1.19-3, but this may not be > for the reason I stated. Any particular reasons for using 1.19-3 ?? There's a problem here anyway... Probably a simplification of the CAS that shouldn't our ==== > It doesn't work correctly in 1.19-3, but this may not be > for the reason I stated. > Any particular reasons for using 1.19-3 ?? > There's a problem here anyway... Probably a simplification of the CAS that shouldn't our Yes. I have no easy aess to upgrades. My old Macintosh won't the right software. I can transfer program files, libraries and such, just not the version upgrades. ==== I'm trying to set up my HP48GX for college math & engineering. When trying to use both Metakern 2.3 & Erable 3.2, there is hardly any ram left, since both seem to need to be in port 0. ( I have both the 128k & 1 Meg cards) Do I need both programs? and if Erable is loaded, is g48 redundant? What is the best combination? The modified equation and matrix editors are great, & MK helps with moving LIB & programs around...still, TIA ==== If you want to have l in port 1, then I'd suggest kicking out MK, and use Filer48 instead (www.hpcc.org). There are some EQ writer and Matrix editor replacements around at www.hpcc.org , which don't take up much space. So it should be possible to have your Erable and the above stuff in port 1 at the same time. > I'm trying to set up my HP48GX for college math & engineering. When > trying to use both Metakern 2.3 & Erable 3.2, there is hardly any > ram left, since both seem to need to be in port 0. ( I have both the > 128k & 1 Meg cards) Do I need both programs? and if Erable is loaded, > is g48 redundant? What is the best combination? The modified > equation and matrix editors are great, & MK helps with moving LIB & > programs around...still, ==== > I'm trying to set up my HP48GX for college math & engineering. When > trying to use both Metakern 2.3 & Erable 3.2, there is hardly any > ram left, since both seem to need to be in port 0. ( I have both the > 128k & 1 Meg cards) Do I need both programs? and if Erable is loaded, > is g48 redundant? What is the best combination? The modified > equation and matrix editors are great, & MK helps with moving LIB & > programs around...still, and Erable in port 17 (the highest port in my 2mb card). There are no libraries at l in port 0, only directories. When I want to use Erable, I copy the main library (lib 788) to port 0. (That's the only part of Erable that has to run in port 0 or 1.) If necessary, I (temporarily) move some of my directories to another port on the 2mb card. This gives me enough room to use Erable. When I'm doing something besides Erable that requires more memory, I just remove lib 788 from port 0 (and move back any directories that I need). In practice I find that keeping both Erable and my most frequently-used directories in port 0 works most of the time, with only an oasion need to move things around. when I need to switch things around. I so use LIBMAN to put as many stand-one programs as possible into libraries. That way I can instl them in a port on the 2mb card instead of keeping them in directories. This reduces the amount of space in port 0 that needs to be located for directories. Not l programs will work this way, but many will. Some programs can be executed directly from a port (without being part of a library), so they can be added to CST and run that way if you don't want to bother creating libraries for them. ==== WOW! , I just downloaded the LIBMAN and File48 we'll see how this goes. >> I'm trying to set up my HP48GX for college math & engineering. When >> trying to use both Metakern 2.3 & Erable 3.2, there is hardly any >> ram left, since both seem to need to be in port 0. ( I have both the >> 128k & 1 Meg cards) Do I need both programs? and if Erable is loaded, >> is g48 redundant? What is the best combination? The modified >> equation and matrix editors are great, & MK helps with moving LIB & >> programs around...still, and Erable in port 17 (the highest port in my 2mb card). There are no >libraries at l in port 0, only directories. When I want to use Erable, >I copy the main library (lib 788) to port 0. (That's the only part of >Erable that has to run in port 0 or 1.) If necessary, I (temporarily) >move some of my directories to another port on the 2mb card. This gives >me enough room to use Erable. When I'm doing something besides Erable >that requires more memory, I just remove lib 788 from port 0 (and move >back any directories that I need). In practice I find that keeping >both Erable and my most frequently-used directories in port 0 works >most of the time, with only an oasion need to move things around. >when I need to switch things around. I so use LIBMAN to put as many stand-one programs as possible into >libraries. That way I can instl them in a port on the 2mb card instead >of keeping them in directories. This reduces the amount of space in port >0 that needs to be located for directories. Not l programs will work >this way, but many will. Some programs can be executed directly from >a port (without being part of a library), so they can be added to CST >and run that way if you don't want to bother creating libraries for them. ==== > and so assumes the Gregorian cendar to begin with, it would > appear that Christian religions dominate the current ISO standards, > which will l change, of course, when the Romulans take over. > What do standards mean, anyway? Do they mean that > this is the only way lowed to express things, > or do they mean this is a popular way to express things, > in some places, but intelligent humans of different cultures > can readily translate between one form of expression and another, > using these 'standards' simply as a guide to their meanings, > and need not fight or kill each other over a phrase. though I'm not a friend of the standarization mania in the European union, the above seems to express the well-known US-tendency to ignore the rest of the world. I even understand this. When I was living for a longer period in Cifornia, it seemed to me that the rest of world is rely sml and unimportant, in particular Germany The most striking proof of this tendency (which joins with a slightly exaggerated understanding of the word freedom) in the software domain is Microsoft's treating of the cendar functions. Do you know how the metric system was introduced in Europe? It's a result of the only 3 years lasting rule of Napoleon on the entire Europe (except England and Russia). A taylor was forbidden to employ anymore the yard (Elle) as a measure, on pentry of death. But people were smart enough after Napoleon was kicked out to keep his reforms in most domains of the life until today. - ==== Regarding the ISO8601 week numeration scheme, it strikes me as very > strange. Does anyone actuly use it? Maybe in Europe? > We do. ==== > Do you know how the metric system was introduced in Europe? > It's a result of the only 3 years lasting rule of Napoleon > on the entire Europe (except England and Russia). > A tailor was forbidden to employ anymore > the yard (Elle) as a measure, on penty of death. Oh, those were the good old days! http://www.poplyrics.net/waiguo/soundtrack/damnyankees/012.htm http://www.stlyrics.com/lyrics/damnyankees/thosewerethegoodolddays.htm -[]- . ==== > and so assumes the Gregorian cendar to begin with, it would > appear that Christian religions dominate the current ISO standards, > which will l change, of course, when the Romulans take over. > What do standards mean, anyway? Do they mean that > this is the only way lowed to express things, > or do they mean this is a popular way to express things, > in some places, but intelligent humans of different cultures > can readily translate between one form of expression and another, > using these 'standards' simply as a guide to their meanings, > and need not fight or kill each other over a phrase. though I'm not a friend of the standarization mania in the European > union, the above seems to express the well-known US-tendency to ignore > the rest of the world. How so? I don't see where is saying that Europe shouldn't use whatever standards it sees fit to use. So long as the definitions of standards provide a way to translate between them (and current U.S. cusary units are defined in terms of SI units), and it's clear which unit is being used, what's the big problem? > I even understand this. When I was living for a > longer period in Cifornia, it seemed to me that the rest of world is > rely sml and unimportant, in particular Germany I might have said: ...in particular France > The most striking proof of this tendency (which joins with a slightly > exaggerated understanding of the word freedom) in the software domain is > Microsoft's treating of the cendar functions. I hope that you aren't saying that Microsoft typifies the U.S.A. Microsoft seems to have a bit of a problem sticking to its own standards, even within the same application, and its business ethics are rly despised, though few doubt its marketing prowess, and its ability to employ the best lawyers that money can buy. But to be sure, in the U.S.A., we usuly settle for things being ok, even if not strictly i.O. For example, if my lawn isn't mown and trimmed quite the way someone else would approve of, well, I'm not charging anyone for the privilege of looking at it, so everyone is entirely free to keep his eyes on the road as he drives by instead of inspecting my property, and should have no cause for complaint. Of course, people who choose to live crowded shoulder to shoulder within a city tend to have a slightly different attitude about what their neighbors do (or fail to do). > Do you know how the metric system was introduced in Europe? It's a > result of the only 3 years lasting rule of Napoleon on the entire Europe > (except England and Russia). Actuly, I thought that it was introduced a bit before he came into power, but certainly he was very influenti in effectively enforcing its use, and over a much wider region than might otherwise have been the case. > A taylor was forbidden to employ anymore > the yard (Elle) as a measure, on pentry of death. I'm pretty sure that both the yard and the ell (1-1/4 yard, used for measuring fabric) are English units. Were these perhaps so non-metric French units? Why was Napoleon forcing anyone to use either of them rather than the glorious French metre? I think that on penty of death so became quite popular in France a bit before Napoleon's time, though perhaps not for that particular offense. Certainly Napoleon wasn't the first to decree extreme penties, some of which may well have been considered to be a fate worse than death, for using fse weights and measures, and in particular for debasing the coin of the rem. > But people were smart enough after Napoleon was kicked out to keep his > reforms in most domains of the life until today. That explains it. Napoleon sold off most of France's claimed territory in the New World and his influence here was minim, though his actions no doubt seemed to raise plenty of cause for concern. Americans, like anyone else, quite naturly resent any attempt to have foreign standards imposed on them, and Napoleon isn't around to force us to adopt them. The standards that we inherited from England with its vast empire have served us quite well; if they need improvement, we could ways do it ourselves; why go looking to those strange Continent standards? == 1900 is not a leap year, as is 2100. 2000 is a leap year. With Excel 2002 I find that: 1900 is considered as a leap year ===> BUG (Microsoft are you reading this). 2000 is considered as a leap year ===> OK 2100 is not considered as a leap year ===> OK. It may be because this software use 1900-01-01 (january first 1900 as its starting date) but to me it is no excuse for poor programming. As a refresher for those who forgot the definition of a leap year here it goes as I used to programmed it in BASIC on my old APPLE II back in 1978 and HP41C/CV/CX: If it is divisible by 4 it is a leap year. If, however, it is so divisible by 100 the it is no more a leap year. If, however, it is so divisible by 400 then it becomes again a leap year. In BASIC I set the number of days for February at 28 then performed three IF THEN tests in suession cooresponding to the three statements above. At the end of these tests, the number of days in February is set to the proper vue. Of course a huge IF THEN ELSE nested test will do the job but is more cumbersome to program and debug. I know, I used that first then simplified it with the three simpler tests. As for testing Excel, here is how I did it: In any cell, type a date (I use the format of YYYY-MM-DD), for example: 2000-02-20 In an adjacent cell (under is easier) type the date for the next day, for example : 2000-02-21 Select the two cells. Extend the selection for at least 10 more cells. Excell creates a serie of dates, switching from 2000-02-29 to 2000-03-01 at the proper place. It will do it correctly if you use 2100 instead of 2000 but will not do it properly when using 1900. I agree that 1900 is in the past and we do not rely need it for present scientific or commerci cculations but why not program it correctly ? Used as an historic helper for date cculations Excel is not perfect. It is better to use your own formulas than to rely on the built in ones if you need to go back in time before 1901. Hoping that this long post helps some of you. ==== > I disagree that work weeks are not used in a significant way. In > virtuly any industry I know of where project plannemanagers and > engineers are used, work weeks are an important part of project > scheduling. Yes, I agree; work weeks are indeed quite often a very important part of project scheduling, and often of manufacturing scheduling. But I've never seen numbered weeks of the year used, but rather week of [date] or week ending [date], or sometimes, particularly when the project is still in the planning phase, week [number within the project schedule]. But perhaps it's done differently in the industries that you're familiar with. > In the semiconductor industry, often times a company will > use work week and year notation (denotingmanufacture date) in the lot > codes for the devices they sell. I'll take your word for it, and of course we know that HP has long used week numbers in it manufacturing codes, at least for the finished product. Perhaps that's typic of Silicon Vley companies. In my person experience, lot numbers more typicly include code for the year, month, and day of month, or the year and day of year, and quite often so the time, or at least the shift. For example, for a lot labeled at 1:37 PM today, a lot number might include a code such as such as 200306031337, 200306031, 20031541337, or 20031541. Some companies seem to prefer obfuscation in their lot numbeso might use 133715403 instead, or completely reverse the digits to come up with 73314513002. I don't know why they bother with such tactics, as the purchasing company rly requires that information on how lot numbers are assigned be provided as a condition of doing business, and could ways just ask when, where, on which machine, and with which tools a particular lot of components was made. Some companies simply use consecutive numbers for lots, so when there's a need to know when a lot was made, they need to dig into their records to find the information. Nowadays, a vendor can expect to be in deep trouble if it can't produce detailed information for traceability purposes quite soon after it's asked for. And the consequences of not being able to narrow things down should a product recl be required ought to be enough to make any vendor shudder. And of course, food, beverages, and drugs have (I expect by law), a pull or sell by: code on the package, after which the product must not be sold, nowadays usuly in a notation that a consumer can easily understand, but sometimes still in an obscure proprietary code that the retailer has to decipher, perhaps so that the consumer won't pick and choose for the freshest product. I've noticed that milk (at least where I live) now includes a time as well, I suppose more for traceability purposes than providing a pull time. Products with a short shelf life, such as some baked goods, often add a use by: date, I suppose to avoid scaring off consumers who notice that the sell by: date is very soon. Some other perishable products, such as batteries, now usuly include a best if used by: date. I've never noticed any week number in any of these codes, though I suppose that it could be hidden within a proprietary code. > I have never tried to understand the method by which the first week is > cculated but I am sure the use of work weeks is wide spread. Well, as long as such numbers are intended to be for intern use only within the company, and not for communications with outsideI don't see that it rely esave that it might make things a bit easier when employees move on to a different company. ==== > and so assumes the Gregorian cendar to begin with, it would > appear that Christian religions dominate the current ISO standards, > which will l change, of course, when the Romulans take over. I did not found Romulans in my dictonary, I think you tk about iens, acting in Enterprise? The important thing for a standard is: It is a *convention (underlined and bold). And we are tking about an internation convention. It is maybe nothing to comunicate with your haircutter, but think about a visa for e.g. belgium which lasts the date for 2003-01-12 which is in januar). Would the visa still vid for you. No, but you would think so. What is more vid than an *internation standard. (Ok, there might me exeptions). ISO8601 has been adopted widly to nation standards, please refer to: http://www.demon.co.uk/datefmts.htm#8601 I canot see any need to invent or use something different for week-numbering or date formats. Defining the begin of a week is partly a question of taste and in case of having different tastes, a convention is required. Non christian religions can benefit from this convention as well. I do not like to discuss about *conventions, which has ready been made. In this case we are tking about an internation convention, which has been adobted e.g. to ANSI... The only problem is, that it is not popular, and nobody is aware of it and only a few companies sees a benefit in using the standard. It would never become popular, when it would not been used. Important standards become typicly to nation *laws, e.g. when there is a benefit for the country, in using it. For internation deing companies, there is a need to do so. The benefit of ISO8601 is avoiding confusion when comunicating word wide. This typicly means adaption for each country and each person. For some people it seem to mean loss of liberty, and my feeling is this is espessily vid for the US , but this would be the price for a more trouble-free comunication. > Sever Microsoft programs (such as Aess 2000 and Excel 2000) Having different standards means: having *no standard. MS is a good example how to make *money, *not how to avoid confusion - Would'nt it be nice to get rid of flag -42? Sticking to obsolete loc conventions is an obstacle for progress. An *old HP advertisment says: Wer nicht mit der Zeit geht, geht mit der Zeit. (Whoever doesn't go with the time goes with the time). *Old means in this case *good. ==== 1900 is not a leap year, as is 2100. 2000 is a leap year. With Excel 2002 I find that: 1900 is considered as a leap year ===> BUG (Microsoft are you reading > this). 2000 is considered as a leap year ===> OK 2100 is not considered as a leap year ===> OK. It may be because this software use 1900-01-01 (january first 1900 as > its starting date) but to me it is no excuse for poor programming. Actuly, I've read that Excel's Day 0 is rely 1900-01-00 (January 0th, 1900). It's a Microsoft product.... The story that I've read is that when they first released Excel, they hadn't yet reized that addition leap year rules apply to years divisible by 100. Isn't every year divisible by 4 a leap year? Hasn't it ways been that way? (Microsoft folks seem to be a little slow in catching on to how outsiders do things, unless perhaps it's something that they can force-fit into their own products, or maybe exploit to make another product incompatible with future Microsoft products.) By the time that they reized the error, quite a few spreadsheets had ready been made with this application. The only year affected by the error in the origin version was 1900. What to do? I suppose that they could've attempted to notify their cusers that there was an error, and provided a method of correcting any spreadsheet that would be affected by it. But that might just possibly have made a lot of cusers a bit unhappy with Microsoft, and would've cost money and wouldn't have been the least bit convenient. Or they could perhaps have made the next version of Excel recognize spreadsheets with the error and auaticly correct it; maybe even sneak it in without even telling the cusers. (Surely l cusers are required to purchase an upgrade As Soon As Possible.) But hey, who cares about the year 1900 anyway? People that ancient must surely have been senile by that time and would never notice a day more or less anyway. So they simply left the error in. Problem? What problem? Any such imagined problem has been solved by simply changing the cendar! After l, if a mere Pope could mandate a modified cendar, surely that power couldn't be denied to a company as powerful and self-important as Microsoft! I do hope that no one misses the sarcasm in the above. present scientific or commerci cculations Unless we're deing with dates in 1900 or earlier. > but why not program it > correctly ? It's sometimes hard to admit a mistake, and Microsoft seems to feel that it can get away with it. Maybe Honest Gates the Wonder Child has a bit of Fonzie in his personity, and can't quite pronounce the wr--g word when referring to himself or his company. > Used as an historic helper for date cculations Excel is not > perfect. It is better to use your own formulas than to rely on the > built in ones if you need to go back in time before 1901. And I wouldn't trust it for various other things without doing some testing first. I've seen some very strange, and sometimes completely nonsensic, statistic results supposedly rated by spreadsheets. But those may been cases of user errors rather than re application bugs. Computers never lie, so it has to be right! Yeah, sure, but programmers and users do make mistakes, and could it be that some suppliers apply a fudge factor when it seems, at first glance, to make the results change to aeptable? Your RPL cculator, or a speciized application, does a better job with these tasks. > Hoping that this long post helps some of you. It satisfied some of my curiosity, and may have saved some Excel users a bit of confusion and trouble. > Jean (ny) Lemire from Richelieu, Quebec, Canada. Ah, the Quebecois rely can use English after l, at least when communicating with non-Canadians. ==== > 1900 is considered as a leap year ===> BUG (Microsoft ... I mentioned ready that a striking example of ignoring established > norms is Microsoft's treating of cendar functions. There are more > mistakes. E.g., Microsoft uses its own simplification of computing the > first work-week in a year. A scand! Hey ? Ah well, I've long since ceased to be shocked by any of the scandous sins that Microsoft commits. They seem to invent their own standards, and apply, ignore, as change them in whatever way seems convenient (or maybe sometimes even inconvenient) to them for a particular product or version. Best to expect that their products won't work quite the way that they ought to. I'm no fan of Microsoft, and I'll very rarely (maybe never?) even attempt to defend their actions. ==== > Microsoft uses its own simplification of computing > the first work-week in a year. A scand! Yes, but they are only three days off! Microsoft's week #1 is the week containing Jan 1, while ISO's week #1 is the week containing Jan 4 (of course, make sure that Microsoft weeks begin on Monday, too :) The Microsoft work day so contains a non-standard 27 hours, but that's an entirely different subject ==== >With Excel 2002 I find that: 1900 is considered as a leap year ===> BUG (Microsoft are you reading >this). 2000 is considered as a leap year ===> OK 2100 is not considered as a leap year ===> OK. It may be because this software use 1900-01-01 (january first 1900 as >its starting date) but to me it is no excuse for poor programming. The bug was in Lotus 123. Excel *deliberately* copied it in order to maintain compatibility. ==== > 1900 is considered as a leap year ===> BUG (Microsoft ... I mentioned ready that a striking example of ignoring established norms is Microsoft's treating of cendar functions. There are more mistakes. E.g., Microsoft uses its own simplification of computing the first work-week in a year. A scand! Hey ? > I agree that 1900 is in the past and we do not rely need it for > present scientific or commerci cculations ... In the contrary. It is very important for historians to have a precise cendar of the past. But don't worry. though I don't like to annouce own tools in the NG, the library Timeman computes the leap years absolutly precise, back to the beginning of the Gregorian cendar and forth to the far future... And the slightly expanded version displays so the week number in aordance with the norm, in both the EU- and the US-cendar. Thus, forget about the Microsoft cendar tools ==== I has done some little programs of topography in USER-RPL. On first versions of them I did not use INPUT command.When executed it got data directly from stack. The trouble were if a long time passed I did not remember the order to enter data. I must view the program to put data on stack in the correct order. It wastes time. Nowaday I use de INPUT command. I find another trouble with this. I must enter data (lists of coordenates of X, reduced distances, azimutes,... This data is stored on the stack (they are the result of previous executed programs) They are there but I can use them with INPUT command. A solution is to enter the level of stack where wished data is. f.e. If I know the list of coordenates is at level 1 of stack I enter 1 PICK instead the whole list. It is ok if you must enter 3 or 4 number, but if I wanna enter more data I do not know what is the level of the stack I must enter. I have saw on some programs On soft-menus appear options. I wanna do the following: I run stack (or something like this) To Correct coordenates propor- With this I wanna switch beetween cionity to distances. INPUT command and the STACK. . Enter Incr X, Incr Y & Dr. :: |go stack|copy stack|paste| Nowadays I use the INPUT command > [and want to use it to choose input from the stack?] As INPUT is being executed in the HP49, you can just press HIST to display the stack and choose something from it; when you make your choice and press ECHO, whatever line you have chosen is inserted into the text being edited by INPUT, at the cursor's current location. Something similar can be done on the 48G, by doing 28 MENU before INPUT, which displays an editing menu with the stack browsing label above the F key (28 MENU so works on the 49G, but shows the editing TOOL menu). One trick with INPUT is to follow it with { SWAP + before OBJ-> With this trick, you can be most certain that the fin result will be a single list of zero or more items, depending on what you left in the edit line, and you can then determine how many objects got created by doing SIZE or OBJ-> on that list; this so prevents any commands typed into the edit line from being inadvertently executed. The CHOOSE command is an ternative to INPUT, when l that you want to do is select one item from a list. Another possibility is the PROMPT command, which displays a message, but leaves you able to use the stack any way you wish; when you are ready to continue, just invoke CONT (either from the keyboard or from a menu which you have ready displayed). A slight extension of PROMPT is an extern command which we cled INMENU, which not only gives a prompt, but so takes a list of variable names which then appear in a menu, so that to store anything into your various named variables, you can just press the corresponding keys, and when you are finished storing vues, in whatever sequence you want, you can then press an OK key to continue your program. The program for INMENU was posted here: http://groups.google.com/groups?selm=3DA3BC89.465907D0%40miu.edu The above version of INMENU has a built-in default prompt, but it can be modified to aept a user string instead, by simply omitting the built-in default prompt string. There's apparently a little INPUT bug in 49G version 1.19-6: If you have used ECHO during INPUT, then cancel INPUT via ON ON, the last arguments to INPUT will not be returned to the stack (whereas on the 48G the args are returned as expected when flag -55 is clear). This can mess up 49G programs which use IFERR and expect the arguments of INPUT to be returned if INPUT is canceled, as they normly would be if only ECHO had not been used. . ==== > There's apparently a little INPUT bug in 49G version 1.19-6: If you have used ECHO during INPUT, then cancel INPUT via ON ON, > the last arguments to INPUT will not be returned to the stack > (whereas on the 48G the args are returned as expected > when flag -55 is clear). This can mess up 49G programs which use IFERR and expect > the arguments of INPUT to be returned if INPUT is canceled, > as they normly would be if only ECHO had not been used. Actuly, it's using HIST at l which causes it, Yes, the first thing HIST does is CK0 which will save the LASTARG. So should an error our the origin stack for the input will not be restored. This is not the only time it could happen. While in input mode, goes into HT then CONT. You'll see that the last arg has been lost. Unfortunately nothing much can be done there. Starting any user RPN command while in USER MODE (with the program starting with TakeOver) would so destroy the last arg for INPUT... I don't consider this as a bug. You gain a feature in the HP49 as side effect it doesn't entirely behave like the HP48 anymore ==== > Unfortunately nothing much can be done there. Just as I've often said myself, the day before someone else came up with a solution to my problem :) > I don't consider this as a bug. You gain a feature in the HP49 > as side effect it doesn't entirely behave like the HP48 anymore. I don't quite see what feature has been gained in the 49G over the 48G with respect to the INPUT command; both cculators identicly permit using the Stack Browser (a/k/a Interactive Stack) to possibly echo decompilations of stack objects into the text being edited. However, in the 48G this never wipes out last args, whereas they are definitely wiped out in the 49G, causing any program like -55 CF ... IFERR INPUT THEN ... END to be inconsistent in whether the args to INPUT are returned, depending on whether or not the stack history was looked into during the INPUT command. Perhaps the HIST key could be more careful while in editing mode, as evidently was the stack browser menu key definition in the 48G (in menu 28). It's a sml thing, but illustrates how even the smlest changes can have unexpected repercussions throughout wider areas of a system: http://www.me.gatech.edu/me/publicat/AugTranscript.htm I do hope that every engineer's education includes an outstanding presentation like that gem. ==== you can find copyright info here: http://lcweb.loc.gov/copyright/ The link for copyright basics would suggest that mathematics as Ideas, procedures, methods cannot be copyrighted but a description/explanation could be (like in a book). so if you use another explanation (changing the words around, explaining the same idea), then you should be ok. Mathematics in r can be considered common property i think. from the site copyright basics link: WHAT IS NOT PROTECTED BY COPYRIGHT? Sever categories of materi are rly not eligible for feder copyright protection. These include among others: *Ideas, procedures, methods, systems, processes, concepts, principles, discoveries, or devices, as distinguished from a description, explanation, or illustration *Works consisting entirely of information that is common property and containing no origin authorship (for example: standard cendaheight and weight charts, tape measures and ruleand lists or tables taken from public documents or other common sources) *Works that have not been fixed in a tangible form of expression (for example, choreographic works that have not been notated or recorded, or improvisation speeches or performances that have not been written or recorded) *Titles, names, short phrases, and slogans; familiar symbols or designs; mere variations of typographic ornamentation, lettering, or coloring; mere listings of ingredients or contents > I guess I could be a little paranoid. I'm a musician, and in music, > there are a million ways of violating someone's copyright. I remember > my music theory teacher in high school mentioning that when he was > copying staff paper, he had to make sure to copy the copyright as > well. The only reason he was lowed to copy staff paper at l was > because he was a teacher. For those that don't know, staff paper is > BLANK music paper. There are (aording to my music teacher) > copyright guidelines for blank music paper! I would hate to see someone get involved in a lawsuit for trying to > help people. It'd be even worse if I thought that it might be a > problem and didn't say anything. I guess that we probably don't have > much to worry about though; Metlica isn't writing textbooks as far > as I know. ;) ==== take a look here: fair use rule if the purpose of copying serves the use of scholarship or education, you are lowed to copy un authorised materi in certain circumstances deemed to be fair even if the owner does not give you permission. if you are using a sml portion of someone else work in a non competitive way and to benefit the public, it should be ok. For example, you are free to quote from your research resources. what is fair? use for non competitive purpose. for example another book would be seemed competitive. if it affects the ses of the copied materi then is competitive. how much materi was taken? the less someone takes, the better. how was the materi used? did you add meaning? did you provide new insights? remember that these are a non commerci free cculator activities aimed at teaching the use of a cculator as a tool to solve problems. maybe making an statement at the beginning would help. please try to use your own words and be creative as much as possible. an example of a statement at the beginning would be: this materi was prepared for non profit education use or something of such nature. > you could use problems from text books > that you find interesting, This brings up a question: How conscious do we have to be about copyright laws when we are > posting the answers to problems from our textbooks? I know that what we're thinking of doing is sml, but companies have > freaked out over fairly sml copyright infringements. Would we have to get permission, document our sources, change the > problems so they are no longer someone elses? ==== My main concern was the problems themselves. In a Cculus textbook, the problems are thing like, Differentiate X^2 which of course I wouldn't worry too much about copying word for word. Conversely, in an Engineering textbook, the problems have usuly been researched and specily written (by the author's understudy) to illustrate a certain point. They usuly consist of drawings and a fairly lengthy explanation, and definitely represent someone's work. After thinking about it some more though, if someone posts something that they needed permission to post, the publishing company will most likely just want it removed. Leg action costs them money and makes them look bad. well, i hope nobody copy/paste a problem from a book. one should be able to understand a topic enough to be able to create his own exercise. there is nothing wrong with looking for problems in books to give you ideas for inspiration. i think copyright only applies to form not ideas. maybe people i could send requests of problems they would like to be seen solved with the help of a graphing cculator (which could or could not be aepted). that way we may provide something very vuable. separating a problem's author with a solution's author (or authors) would suggest originity and reity. this is an idea: we could form a team of voluntary users that solve the problems other people send. by breaking down the problems in series of steps solved by a team, then there would be no question at l about the originity. by dividing so we ease the work load. for example one person finds a solution (by any of the team members or combination of them). other person later makes pictures or adds other content. other person would put together l .And other person would manage the site. it would be something like a fast food chain (but without the rush). we could chose form a list of problems people send that would have the most vue. would anybody be interested in being part of a solvedproblems team or suggest other ideas? just an thought ==== Hopefully they would. A posting guide might be a good idea though. I could try to write one up and send it to you if you want. > maybe people i could send requests of problems they would like to be > seen solved with the help of a graphing cculator (which could or > could not be aepted). that way we may provide something very > vuable. separating a problem's author with a solution's author (or > authors) would suggest originity and reity. this is an idea: we > could form a team of voluntary users that solve the problems other > people send. by breaking down the problems in series of steps solved > by a team, then there would be no question at l about the > originity. by dividing so we ease the work load. for example one > person finds a solution (by any of the team members or combination of > them). other person later makes pictures or adds other content. other > person would put together l .And other person would manage the site. > it would be something like a fast food chain (but without the rush). > we could chose form a list of problems people send that would have the > most vue. would anybody be interested in being part of a > solvedproblems team or suggest other ideas? I wouldn't mind being on that team. I'm good with finding solutions for lower level mechanic engineering (first two yeafor now), but not much good with anything that involves running a webpage (html, incorporating drawings, etc.) I can create bitmap drawings using POVRay (its free), but it's not one of my favorite things to do. AutoCAD is easier to use, but the program is expensive and I have ethic problems with the way they do business. We could send each other the solutions that we came up with so the rest of the team could look at it and maybe optimize them (make them shorter, clearer, etc....) That way, we could learn from each other while we're doing this. This will so help with the TI solutions, because I don't own one. I'm around them so much that I (sort of) know how to use them, but I'm sure that someone like knows a whole lot more. Do you suppose we should maybe start gathering a list of people, and way we could take this off the NG. == > Hopefully they would. A posting guide might be a good idea though. I > could try to write one up and send it to you if you want. > I wouldn't mind being on that team. I'm good with finding solutions > for lower level mechanic engineering (first two yeafor now), but > not much good with anything that involves running a webpage (html, > incorporating drawings, etc.) it would be a good idea to focus on easy ones first (like cculus). don't worry about the page, i like to learn new things as i go ong (like php). > I can create bitmap drawings using POVRay (its free), but it's not one > of my favorite things to do. AutoCAD is easier to use, but the > program is expensive and I have ethic problems with the way they do > business. maybe xray tracing is a little too powerful. i think that simple cculator screenies would help the most. i can so do sml things in autocad or electric squematics with simulink (whatever the need may be). > Do you suppose we should maybe start gathering a list of people, and > way we could take this off the NG. web based would be better. we'll see. just remember that we are supposed to have fun ;) ==== it would be a good idea to focus on easy ones first (like cculus). > don't worry about the page, i like to learn new things as i go ong > (like php). web based would be better. we'll see. just remember that we are > supposed to have fun ;) Count me in, I can help with various math, cculus, and engineering problems/topics if and when I have the time. It would be helpful for the webpage to contain a form where people could post their problems and we could introduce guides or solutions to these problems describing procedures using HP and TI ccs. ==== The basic cculus marathon was updated and iploaded to hpcc. This time we have: 1) Examination of the knowledge of the HP49G when it comes to derivation rules. Implementation of programs that *collect* differenti forms, i.e. dx(f)*g+g*dx(f)->d(f*g). (Works so for sums, just read the marathon ) 2) Program for converting form derivatives from the notation d1F(X) to the notation d/dx(F(X)) 4) How to use programs that give themselves arguments that contain... themselves. 5) How to use MATCH to trigger execution of programs. These triggered programs can be so DBUGed. 6) How to use MATCH in combination with the arms, to schedule triggered program execution. 7) How to follow exactly (step by step) what MATCH does. 8) How to use MATCH to implement easily the most wished feature INTEGERASSUME !!! It seems that after so many years MATCH still has a huge potenti. 9) Finding derivatives of inverse functions. 10) Finding derivatives of parametric and polar functions. 11) Example of an most square (!!!) orbit of a satellite of a planet around the star. 12) Example of a simple machine that creates an unexpected group of curves. ...And much more ==== Curiously, even the function numbers in library 171 (l the User commands and functions which were new to the 48G, meaning that they were not in the 48S) changed enroute to the 49G. A number of people have re-released their own libraries with changes in function order (and hence function numbering); this is a bad idea for named functions, because it then requires recompiling l programs which used the commands, but there is nothing sacred against doing this for unnamed intern functions which have never been declared to be supported. It would be nice, nonetheless, for the most useful and important of such intern functions to remain unsupported but stable, but trivi functions er less. Some trivi functions er more, however consider that lists of re numbers may at first guess seem portable in binary form between HP48 and HP49, but a closer look reves that a certain few vues in such lists might be the speci vues abbreviated by ROM addresses, e.g. single-digit re numbers; unfortunately, these ROM addresses are not compatible between 48 and 49, hence lists of res are not necessarily binary compatible. Vectors and matrices of res made on 48 are compatible with 49, however (but not ways the other way around why?) ==== Nobody never promised a binary compatibility! Use ASCII transfer. The UserRPL source works and even most of the SysRPL source works. You should *ways* recompile !! > Curiously, even the function numbers in library 171 > (l the User commands and functions which were new to the 48G, > meaning that they were not in the 48S) changed enroute to the 49G. A number of people have re-released their own libraries > with changes in function order (and hence function numbering); > this is a bad idea for named functions, because it then > requires recompiling l programs which used the commands, > but there is nothing sacred against doing this for unnamed intern > functions which have never been declared to be supported. It would be nice, nonetheless, for the most useful and important > of such intern functions to remain unsupported but stable, > but trivi functions er less. Some trivi functions er more, however consider that > lists of re numbers may at first guess seem portable in binary > form between HP48 and HP49, but a closer look reves that > a certain few vues in such lists might be the speci > vues abbreviated by ROM addresses, e.g. single-digit > re numbers; unfortunately, these ROM addresses > are not compatible between 48 and 49, > hence lists of res are not necessarily binary compatible. Vectors and matrices of res made on 48 are compatible > with 49, however (but not ways the other way around why?) ==== > Nobody never promised binary compatibility! On the contrary, most data objects (complex numbesystem or user binary integers, graphic objects, re or complex vectors and matrices, *most* re numbeetc. and lists of same) are binary compatible 48 -> 49, because the object type addresses remain the same, as well as the structure. Except mainly for a certain few common res, which are auaticly abbreviated by ROM addresses, even in lists (but not in arrays), and where these few addresses did move. You can change a list of res to a vector of res, however, via LIST-> ->ARRY, and then it can be transferred in binary (with the help of OBJFIX or FIXOB, or a PC hex editor, or an early version of Emu48 with 49 support). Not unit objects either, I expect (but am too lazy to look up :) . ==== > for the hint ot LofRP (Libman) and the TICKS-Pointers. Heiko, > I do not know whether the order of rompointers in lib 181 was changed > it the past. At least, it will not change from ROM 19-6 to ROM 19-7, > unless somebody discovers still a bug in some basic browsers like the > arm or finance box and this is rather unlikely As a er of fact, libraries like Libman or Timeman use essentily > the new features of EXTPRG, introduced by JYA only in ROM 19-6. Thus, > I usuly write in my tool-docus: you should upgrade to ROM 19-6. so > Keyman can be used only after ROM 19-3 when shift-hold was introduced > and the UserKeys list in the Hidden dir changed its structure. It is an illusion to believe that l programs or advance tools are > backward compatible. Many advanced tools written in the period after ROM > 19-6 aren't backward compatible with the supported ROM 1.1-18. IMHO, ROM 19-6 made a decisive step in the developement, not only in > adding new features. Backward compatibility is not as important an issue > as it was in the times of the 48 when upgrading meant to buy a new cc. > Everybody can upgrade in 20 minutes now, with no addition costs. Agreed; there is some vue in writing things that will work on the 48 series cculators as well as on the 49G, so long as source code is provided of course. But if an application is for the 49G only, I see no good reason to make it backward compatible with older ROM revisions. Simply include a statement as to which ROM revision it was written for and tested on. But there is vue in writing in such a way that the application can reasonably be expected to work on any future ROM revisions. > The > only serious problem is updating the documentation... I don't know whether this problem will ever be solved. It c a n be > solved, but only if somebody is willing to pay for it. There are only > few experts good enough for this task. I know only one (in clear and > easily understandable English writing) poster in this NG who I guess > could do the job perfectly. It would be helpful if the documentation were released under some sort of open source license. Then new commands could be added, the Computer gebra Commands combined with the Other Commands in phabetic order, corrections made, better explanations and more examples added, a Kermit ASCII translation table added, and perhaps chapters from the 48S or 48G (updated for the 49G) documentation added. A project to do such things could be started without the risk of violating HP's intellectu property rights. Perhaps urls to other resources such as Eduardo Kinowski's pages at: http://membres.lycos.fr/ekin/index.php would be more appropriate for some things. ====