A3 I have a total of 9 registers for storing: RI and R0 to R8... ... so I have a total of 10 registers for storing. ==== I have an HP 11C and the guidebook says for that error: Named storage register currently converted to program memory, or nonexistent storage register. HTH > My 15C is doing something strange. When I try to store into memory (for > example- sto 4), I get an Error 3. I can only store when I use sto 1. > Every other memory gives Error 3. Anyone know why. I can't find my manual. Chris ==== What is the maximum memory HP 48GX can have? How would one go about 1. obtaining the memory cards/hardware necessary to upgrade? 2. obtaining information about how to upgrade? thanks. louis ==== >What is the maximum memory HP 48GX can have? How would one go about 1. obtaining the memory cards/hardware necessary to upgrade? 2. obtaining information about how to upgrade? thanks. louis of error in the GX addressing the last two banks of a 4 Mb card, the klotz cards permit bank substitution of the last 2 ports. Unfortunately, the extra logic comes at a price since the 4M card is more than twice the cost of a 2M card. I bought the 2M, and it is more than enough. To upgrade, buy one 128k card for port 1, and 1 ?M card for port 2. Plug them into the appropriate ports with the power off. Turn on the calc, and you are all set. PINIT seems not to be necessary as each card has been formatted when tested at the factory. Hope this helps. Bill Bill alternate E-dress wtstorey@ieee.org.no.spam.please (Use the obvious) ==== Am i misreading the price or is it only $70 for a 2 MB card!? -louis ==== > Am i misreading the price or is it only $70 for a 2 MB card!? You're correct. I just ordered a 2 MB card from Klotz a couple of days ago, and the price, including shipping, was about $78. -- Wayne Brown | When your tail's in a crack, you improvise fwbrown@bellsouth.net | if you're good enough. Otherwise you give | your pelt to the trapper. e^(i*pi) = -1 -- Euler | -- John Myers Myers, Silverlock ==== HP sells 1MB, max is 4MB, but some last ports have difficulties in access.Anyway 3,5MB is a lot, too. Remember that a port is only 128KB > What is the maximum memory HP 48GX can have? How would one go about 1. obtaining the memory cards/hardware necessary to upgrade? 2. obtaining information about how to upgrade? thanks. louis ==== > What is the maximum memory HP 48GX can have? How would one go about 1. obtaining the memory cards/hardware necessary to upgrade? 2. obtaining information about how to upgrade? > 1) http://uuhome.de/oklotz/index_e.html 2) http://uuhome.de/oklotz/hp2048ae.pdf ==== I want to convert from rectangular to polar on the 49G. I have tried to do this using the vector notation on the calculator, but when I go to convert the vector (polar) back to its components it flips back to rectangular. I want to do this using a program, so there must be a function to do so. MC ==== > I want to convert from rectangular to polar on the 49G. In the days of much older calculators, there were no complex number objects or vector objects; people had to enter two completely separate real numbers onto the stack and mentally imagine that these were related, which is what gave rise to the need to replace one pair of numbers with another pair from time to time, which was called converting R<->P; the calculator, meanwhile, never knew what was going on, so the user had to take all responsibility for invoking R<->P himself whenever required, and then still dealing separately with each component during other subsequent computations. The modern HP48/49 has internal complex number objects and vector objects; these single objects have only one internal form, which is never converted at all; instead, any conversion occurs automatically during input parsing and output displaying, so that the user need never explicitly convert anything himself. On input, the user enters an angle symbol <) just preceding each coordinate which is an angle, and the calculator immediately and automatically converts that input into its internal form, using the syntax to automatically understand the type of coordinates which were entered: [ x y ] or ( x y ) rectangular [ r <)theta ] or ( r <)theta ) polar [ x y z ] rectangular [ r <)theta z ] cylindrical [ r <)theta <)phi ] spherical For output (display), the user can choose RECT, CYLIN or SPHERE display modes, and the calculator then immediately and automatically converts every complex number or vector (2D and 3D) from its internal form back to the coordinates that the user wants to see (above). All angles, of course, are in the current angle display mode (DEG, RAD, or GRAD), which is one more thing to be set before entering input or interpreting output. Whatever we now want to do with this calculator, the above understandings should now make it clear how to do it. If we want to enter some values in rectangular form and see the same complex numbers or vectors displayed in polar form, we just perform CYLIN or SPHERE once to set the output display style, then start entering coordinates without any angle symbols; each time we press ENTER, we instantly see our input converted from rectangular input to polar output on the display, without having to invoke any explicit conversion functions. If we want to enter some values in polar form and see the same objects displayed in rectangular form, we just perform RECT once; as we enter each polar vector with an angle symbol and then press ENTER, we instantly see it re-displayed in rectangular form, again without having to invoke any explicit conversion functions. Summary: o Input: Enter objects with an angle symbol preceding each angle value; all conversion to internal form is automatic. o Output: Perform RECT/CYLIN/SPHERE to re-display output in desired forms; all conversion from internal form to display is automatic. o By entering in one format and displaying in another, all conversion is automatic. o Internal computations with complex numbers and vectors are always automatically correct, because there exists only one universal internal form for each object type; there is no internal conversion between forms. Other commands which operate on complex numbers: o RE gives x, IM gives y, C->R or OBJ-> give both x and y o ABS gives r, ARG gives theta [ABS also works with vectors] > when I go to convert the vector (polar) back to its components > it flips back to rectangular. Nothing flips back; vector objects exist internally only in rectangular component form, so commands such as OBJ-> and ARRY-> (or the Matrix Writer) will discover only rectangular components inside a vector. HOWEVER, there are three special commands (V->, ->V2 and ->V3) which actually convert between internal (rectangular) components and external (display form) coordinates, depending on the current coordinate display mode (RECT/CYLIN/SPHERE): V-> or a 2- or 3-component vector into the separate values that would be *displayed* in the current mode (RECT/CYLIN/SPHERE). By the way, this is why V-> is not a good idea as a shortcut for ARRY-> DROP, because sometimes it may produce converted, angular coordinates, instead of the desired original components. ->V2 and ->V3 These do the inverse of V-> In other words, in CYLIN/SPHERE modes, these assemble polar coordinates into a vector, just as if you had typed angle symbols in front of those components, thus converting external polar coordinates into internal rectangular components. You can have ->V2 produce either a vector or a complex number object, according to whether flag -19 is clear or set. More rantings, old P<->R revived on HP48/49 for diehards, RAD and POLAR toggles revived on HP49 for those who miss them, angle symbol on a more convenient key, etc. http://groups.google.com/groups?selm=3CA74D8B.55348AC5%40miu.edu http://groups.google.com/groups?selm=3D46423F.6170959A%40miu.edu http://groups.google.com/groups?selm=3B234542.CB2402A%40miu.edu http://groups.google.com/groups?selm=3B230DDB.942573D%40miu.edu With best wishes from http://www.mum.edu . ==== the two. The main problem is that when you do a obj-> command on top of a polar vector it will reduce back to the rectangular form. I want to be able to get the primaries in polar form on two levels of the stack. MC I want to convert from rectangular to polar on the 49G. In the days of much older calculators, there were > no complex number objects or vector objects; people had to > enter two completely separate real numbers onto the stack and > mentally imagine that these were related, which is what gave rise > to the need to replace one pair of numbers with another pair > from time to time, which was called converting R<->P; > the calculator, meanwhile, never knew what was going on, > so the user had to take all responsibility for > invoking R<->P himself whenever required, > and then still dealing separately with each component > during other subsequent computations. The modern HP48/49 has internal complex number objects > and vector objects; these single objects have only one > internal form, which is never converted at all; > instead, any conversion occurs automatically > during input parsing and output displaying, > so that the user need never explicitly convert anything himself. On input, the user enters an angle symbol <) > just preceding each coordinate which is an angle, > and the calculator immediately and automatically > converts that input into its internal form, > using the syntax to automatically understand > the type of coordinates which were entered: [ x y ] or ( x y ) rectangular > [ r <)theta ] or ( r <)theta ) polar > [ x y z ] rectangular > [ r <)theta z ] cylindrical > [ r <)theta <)phi ] spherical For output (display), the user can choose RECT, CYLIN or SPHERE > display modes, and the calculator then immediately and automatically > converts every complex number or vector (2D and 3D) from its internal > form back to the coordinates that the user wants to see (above). All angles, of course, are in the current angle display mode > (DEG, RAD, or GRAD), which is one more thing to be set > before entering input or interpreting output. > Whatever we now want to do with this calculator, > the above understandings should now make it clear how to do it. If we want to enter some values in rectangular form and see > the same complex numbers or vectors displayed in polar form, we > just perform CYLIN or SPHERE once to set the output display style, > then start entering coordinates without any angle symbols; > each time we press ENTER, we instantly see our input converted > from rectangular input to polar output on the display, > without having to invoke any explicit conversion functions. If we want to enter some values in polar form and see the same > objects displayed in rectangular form, we just perform RECT once; > as we enter each polar vector with an angle symbol and then > press ENTER, we instantly see it re-displayed in rectangular form, > again without having to invoke any explicit conversion functions. > Summary: o Input: > Enter objects with an angle symbol preceding each angle value; > all conversion to internal form is automatic. o Output: > Perform RECT/CYLIN/SPHERE to re-display output in desired forms; > all conversion from internal form to display is automatic. o By entering in one format and displaying in another, > all conversion is automatic. o Internal computations with complex numbers and vectors > are always automatically correct, because there exists > only one universal internal form for each object type; > there is no internal conversion between forms. > Other commands which operate on complex numbers: o RE gives x, IM gives y, C->R or OBJ-> give both x and y > o ABS gives r, ARG gives theta [ABS also works with vectors] > when I go to convert the vector (polar) back to its components > it flips back to rectangular. Nothing flips back; vector objects exist internally only in > rectangular component form, so commands such as OBJ-> and ARRY- (or the Matrix Writer) will discover only rectangular components > inside a vector. HOWEVER, there are three special commands (V->, ->V2 and ->V3) > which actually convert between internal (rectangular) components > and external (display form) coordinates, depending on the current > coordinate display mode (RECT/CYLIN/SPHERE): V- > or a 2- or 3-component vector into the separate values > that would be *displayed* in the current mode (RECT/CYLIN/SPHERE). By the way, this is why V-> is not a good idea as a shortcut > for ARRY-> DROP, because sometimes it may produce converted, > angular coordinates, instead of the desired original components. > ->V2 and ->V3 These do the inverse of V- > In other words, in CYLIN/SPHERE modes, these assemble > polar coordinates into a vector, just as if you had > typed angle symbols in front of those components, > thus converting external polar coordinates > into internal rectangular components. You can have ->V2 produce either a vector or a complex number > object, according to whether flag -19 is clear or set. > More rantings, old P<->R revived on HP48/49 for diehards, > RAD and POLAR toggles revived on HP49 for those who miss them, > angle symbol on a more convenient key, etc. http://groups.google.com/groups?selm=3CA74D8B.55348AC5%40miu.edu > http://groups.google.com/groups?selm=3D46423F.6170959A%40miu.edu > http://groups.google.com/groups?selm=3B234542.CB2402A%40miu.edu > http://groups.google.com/groups?selm=3B230DDB.942573D%40miu.edu With best wishes from http://www.mum.edu > . ==== > The main problem is that when you do a obj-> command > to a polar vector it will reduce back to the rectangular form. Do V-> instead of OBJ-> There is no polar vector internally; there are only different ways of entering the data to create the vector, and different ways of displaying the vector on the stack, but the internal components, revealed by OBJ-> are always rectangular. If you are displaying in CYLIN/SPHERE mode, showing angle(s), then V-> will result in the separate *displayable* values, actually converting from the internal rectangular form, which is what you are looking for in this case. Likewise, ->V2 and ->V3 convert displayable separate inputs back to internal rectangular form. -[]- . ==== You probably figured out by now: DUP RE SWAP IM does the trick for you! ==== You know the math, you can program the functions yourself. It is also possible to use SysRPL and dig the old functions from the ROM (The HP 28 had these functions) BUT Currently the POLAR RECT is a display-only automatic conversion. [MODE] Coord will let you choose it or you can use RECT POLAR keywords also you can change the flags directly, if you wish. The internal representation is always rectangular. You can input in either rectangular or polar and the display will show it in your currently selected format. I hope this helped a little. Veli-Pekka > I want to convert from rectangular to polar on the 49G. I have tried to do > this using the vector notation on the calculator, but when I go to convert > the vector (polar) back to its components it flips back to rectangular. I > want to do this using a program, so there must be a function to do so. > MC ==== I've found some situations in integrations when you get the same answer back as the original problem except the HP surrounds it by INT(fx(Xt), Xt, X). Why does the HP do this and what does it mean? I received the info on how to handle this in one specific problem but don't understand what it means or why it does it so I can generalize the information. Carmen ==== > I've found some situations in integrations when you get the same answer back > as the original problem except the HP surrounds it by INT(fx(Xt), Xt, X). > Why does the HP do this and what does it mean? I received the info on how > to handle this in one specific problem but don't understand what it means or > why it does it so I can generalize the information. > Carmen > > The usual indefinite integral operation is to take a function f(x) and return a function (unique up to additive constant), say g(x), whose derivative is the original function. Thus the result must itself be a function of the original variable. But to specify the function to be integrated in the HP , one must specify a variable and a formula for the result bsed on that variable x -> f(x). The f(Xt) *and* Xt are needed to determine the integrand function, Xt -> f(Xt), and the X is needed as argument of the function resulting from the integration. For a given function f(X), 'INT(f(Xt),Xt,X)' represents a function of X but different from f(X). Note that expressions like 'INT(X^2,X,X)' giving 'X^3/3' and 'INT(X^2,X,2)' giving '8/3' evaluate as expected, but 'INT(X^2,2,2)' gives an error message, since there is no *variable* of integration. Expressions like 'INT(SIN(Xt),Xt,ARCCOS(1/2))' and 'INT(Xt^3,Xt,2)' now have perfectly legitimate meanings. ==== Virgil that's what I don't understand. When I put in a function f of x Why does the 49 return f of Xt? It makes no sense to me. If I try integrating what it returns to me I get f of Xtt? I do know what indefinate integration it but I can't understand what the HP 49 is doing here. It behavior seems arcane in many circumstances. Carmen I've found some situations in integrations when you get the same answer back > as the original problem except the HP surrounds it by INT(fx(Xt), Xt, X). > Why does the HP do this and what does it mean? I received the info on how > to handle this in one specific problem but don't understand what it means or > why it does it so I can generalize the information. > Carmen The usual indefinite integral operation is to take a function f(x) > and return a function (unique up to additive constant), say g(x), > whose derivative is the original function. Thus the result must > itself be a function of the original variable. But to specify the > function to be integrated in the HP , one must specify a variable > and a formula for the result bsed on that variable x -> f(x). The f(Xt) *and* Xt are needed to determine the integrand function, > Xt -> f(Xt), and the X is needed as argument of the function > resulting from the integration. For a given function f(X), 'INT(f(Xt),Xt,X)' represents a function > of X but different from f(X). > Note that expressions like 'INT(X^2,X,X)' giving 'X^3/3' and > 'INT(X^2,X,2)' giving '8/3' evaluate as expected, but > 'INT(X^2,2,2)' gives an error message, since there is no *variable* > of integration. Expressions like 'INT(SIN(Xt),Xt,ARCCOS(1/2))' and 'INT(Xt^3,Xt,2)' > now have perfectly legitimate meanings. ==== > Virgil that's what I don't understand. When I put in a function f of x Why > does the 49 return f of Xt? Because it doesn't know how to intgrate the function you gave it. ==== Bless you steen and virgil! I understand! That is the obvious but I wasn't sure and assuming has caused so many problems in the past I didn't want to. So I have to manipulate the equation in whatever many ways to find an expression it does know how to integrate! Carmen > Virgil that's what I don't understand. When I put in a function f of x > Why > does the 49 return f of Xt? Because it doesn't know how to intgrate the function you gave it. ==== > Virgil that's what I don't understand. When I put in a function f of x Why > does the 49 return f of Xt? If you put 'F(X)' on the stack an use the INTVX command you will get 'INT(F(Xt),Xt,X)'. The 'Xt' is what is called a dummy variable, meaning that replacing it by any other variable not already in the expression does not change the meaning or value of the expression. Thus 'INT(F(Xt),Xt,X)' and 'INT(F(Y),Y,X)' and 'INT(F(q),q,X)' all mean the same thing. Good mathematical practice says that dummy variables should not be the same as actual variables. Since the X in 'INT(F(Xt),Xt,X)' is an actual variable, one should avoid 'INT(F(X),X,X), even though the HP49 can deal with it correctly. 'HP does this by appending a t to whatever actual variable is used when required to invent a dummy variable. ==== I was looking around, checking out what's being sold these days, when I ran across this auction: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=1383400767 It looks like a regular 11C but the gold key functions are completely different! The FURLONGS, TIME and LENGHTS functions are particularly fascinating. Was this a custom unit with built-in programs for handicapping horse races? (I know nothing about horse racing or betting.) Anyone know anything about this one??? - Michael -- ==== answered that he didn't know he had a strange 11C, and put it off the site this morning... RC Michael F. Coyle a .8ecrit dans le message > I was looking around, checking out what's being sold these days, when I ran > across this auction: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=1383400767 It looks like a regular 11C but the gold key functions are completely > different! The FURLONGS, TIME and LENGHTS functions are particularly > fascinating. Was this a custom unit with built-in programs for handicapping > horse races? (I know nothing about horse racing or betting.) Anyone know anything about this one??? - Michael > -- > ==== Michael F. Coyle escribi.97 en el mensaje http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=1383400767 Anyone know anything about this one??? > The seller has no idea (and no docs) about this no standard hp 11c... ==== This looks like a bastardized 11C more than anything else. I wonder if the bidders even realize it's not a standard 11C. I certainly would not want it. > I was looking around, checking out what's being sold these days, when I ran > across this auction: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=1383400767 It looks like a regular 11C but the gold key functions are completely > different! The FURLONGS, TIME and LENGHTS functions are particularly > fascinating. Was this a custom unit with built-in programs for handicapping > horse races? (I know nothing about horse racing or betting.) Anyone know anything about this one??? - Michael > -- > ==== It looks like a special purpose keyboard overlay glued above the original one. Maybe for a game program for 2 players. Raymond Michael F. Coyle schrieb im Newsbeitrag > I was looking around, checking out what's being sold these days, when I ran > across this auction: http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=1383400767 It looks like a regular 11C but the gold key functions are completely > different! The FURLONGS, TIME and LENGHTS functions are particularly > fascinating. Was this a custom unit with built-in programs for handicapping > horse races? (I know nothing about horse racing or betting.) Anyone know anything about this one??? - Michael > -- > ==== Given a simple equation like NEW=OLD*(1.+CHG/100.)^PD when I use ISOL to solve for CHG I get On the 48gx CHG=((NEW/OLD)^INV(PD)-1)*100 On the 49g CHG=100.*EXP(-Ln(OLD/NEW)/PD)-100. Both of these yield the same result, I'm just wondering why they use different approaches. Also, the 49g gives me the error ISOL Error: Non unary operator if I have Approx checked in the CAS Modes. If I uncheck it (turn approx. mode off) then execute ISOL, the 49g prompts me to turn Approx. mode on. If I choose YES, it works (returns the expression above). If I try it again (without unchecking Approx mode), it fails with the Non unary operator error. To get it to work again I have to turn off Approx mode, and let it prompt me to turn it on again. Very weird, eh? Am I doing something wrong, or is there a problem with ISOL on the 49g? Steve ==== > Given a simple equation like > NEW=OLD*(1.+CHG/100.)^PD > when I use ISOL to solve for CHG ... And the same with SOLVE in place of ISOL, although I failed to find where these converge in ROM. Since SOLVE is a superset (in the CAS) of ISOL, I don't see why the original ISOL needed to be messed with at all. It can be said that the complete CAS adds much more overall capability, and that discarding some original symbolic functions might have been thought desirable to make room for it, but since the replaced functions turned out to be incompatible (sometimes even not working at all), it's too bad that it also trampled over certain nice things that were just fine and more useful as they were. -[]- . ==== > Am I doing something wrong, or is there a problem with ISOL on the > 49g? The HP49G is not backward compatible with the HP48 series. Even though some commands accidently share names on the two calculators, their do not work alike. ==== > Am I doing something wrong, or is there a problem with ISOL on the > 49g? > The HP49G is not backward compatible with the HP48 series. Even though some > commands accidently share names on the two calculators, their do not work > alike. This also concerns a lot of SysRPL-commands. For instance, the basic commands XYGROBDISP and INNERCOMP. The latter doesn't crash or error anymore but yields the count ONE for *any* non-composite object (somtimes useful). The former distroys the display in suspended state or if directly executed, provided the inserted grob is too large. In the whole, even experts have problems in porting complex SysRPL-programs from or to a HP48 :-) - Wolfgang ==== Just thought I'd let you know how I 'improved' my Urroz books. After I got them I've been using them a lot with the Calculator. Unfortunately they don't stay open so I had to hold them open with one hand. Working the calc with one hand and holding a book didn't really work so I finally stretched them open and the glue from the spine released and pages started falling out ;/ This wasn't good at all so I took them to Kinko's had the spine cut off and the books coil bound. They now lay perfectly flat without an extra hand and I can even fold them back on themselves perfectly! This was great and cost less than 10 bucks a volume! Really terrific, except for the first volume they cut a little too close, but they will be buying me a new one of those! So if you are having trouble juggling the Urroz books (or any book for that matter), this might work for you! Carmen ==== I'm looking for HP-10C and HP-15C owner's manual (in english or, preferably, RC ==== check on Ebay www.ebay.com ==== LONGFLOAT v. 2.0 beta I've made a multiple precision library for HP49 with + - * / sqrt y^x exp ln sin cos tan asin acos atan sinh cosh tanh and some auxiliary functions for both real and complex input. The precision is set in local/global variable DIGITS. Results should be accurate within +/- 1 in last digit. Integers, longfloats and long complex may generally be mixed. The library is to a great extent based on modified algorithms of ZMLIB ( a fortran package for multiple precision). I have done some testing, but could really need some more beta-testers, since this kind of library should be extensively tested. Those who are interested and have time to help please let me know. Gjermund Skailand Things not working: automatic list processing ( but DOLIST/DOSUBS ok) error messages ok but give e.g. XLIB 688 69 Error: Bad argument type ( I haven't really understood how Werner Huysegoms fixed this in his library ;-( Some timings (depends on actual numbers): DIGITS=100, x=sqrt(pi) sqrt(x) 0.65 sec. ln(x) 12 sec. exp(x) 5.3 sec sin(x) 5 sec ( sin(asin(sqrt(2.)/2))^6- 0.125 = ~2.E-100 15.0sec z = (123456789 , 0.123456789 ) sqrt(z) 2 sec ln(z) 15.8 sec exp (ln(z)) - z = ~ (-3.E-91, 4.E-96) 24.5sec atan(z) 7.0 sec DIGITS = 1000, pi precomputed Sqrt(pi) 28 sec pi - (sqrt(pi)^2) ~ 1E-998 57.7 sec Integers: x = 2^330 y = 2^3330 Zsqrt( x ) -> TRUE .34 sec in exaxt mode Zsqrt( x ) -> FALSE .31 sec in approx mode Zsqrt( y ) -> TRUE 11.4 sec in exaxt mode Zsqrt( y ) -> FALSE 7.5 sec in approx mode Let n = Zsqrt(m) EXACT mode: always returns n such that n^2 <= m< (n+1)^2 . Also flag= 1 if m=n^2t. (In accordance with JMH's specifications ;-) Approx mode: returns n such that ABS(m-n^2) is minimum and regardless of result also return flag=0. PS Some constants, ln(2), ln(10) have been hardcoded with 130 digits, if greater precision are required than about 110 digits, these constants will be recalculated and running time will be thus be significantly greater... ==== > LONGFLOAT v. 2.0 beta > <... Some timings (depends on actual numbers): DIGITS=100, > x=sqrt(pi) > sqrt(x) 0.65 sec. > ln(x) 12 sec. > exp(x) 5.3 sec > sin(x) 5 sec > ( sin(asin(sqrt(2.)/2))^6- 0.125 = ~2.E-100 15.0sec Seems rather slow :-( How about multiplication of 100-digit real floats? -- Bhuvanesh ==== <662e00ed.0209300736.607dc17a@posting.google.com>... > > LONGFLOAT v. 2.0 beta > > Seems rather slow :-( How about multiplication of 100-digit real floats? > let x = pi as 100-digit real float with user funtion : x*x 0.257 sec with library internal function : x*x 0.207 sec multiplication of 100 digits integer using internal flashpointer ^QMul 0.189 sec , of course for floating point calculation it is a waste to calculate result to double precision, then throw away half of the digits, but still I don't think there much potensial for improvement :( -Gjermund > -- > Bhuvanesh > ==== > automatic list processing ( but DOLIST/DOSUBS ok) > error messages ok but give e.g. XLIB 688 69 Error: Bad argument type > ( I haven't really understood how Werner Huysegoms fixed this in his > library ;-( If you mean the messages, that's the easy part? Just make sure every command in your lib starts with :: CK0, CK1&Dispatch, CK2&Dispatch .. CK5&Dispatch. Then, when an error occurs and you do eg #68801 ERROROUT, the message will show up with the command name prepended. The Automatic List processing is a bit more difficult, especially since our friends at ACO had changed it, and made a slight error IMHO. (I mean the {1 2 } 3 * -> {3. 6.} part) I have chosen to correct that behaviour in my lib, rather than make the same error. How does it work.. (from memory, so probably not 100% foolproof:-) - in the 48GX, the development team changed the code of the CK&DISPATCH1 routine so that if there was no match, the argument(s) were re-examined to see if either one was a list, and if so, executed the command for every pair of arguments in the list(s). - in the '49, a new object type is introduced: Long Integer. To avoid having to rewrite the dispatching of every command, ACO chose to change the definition of CK&DISPATCH1 again: if no match was found, and the input was a Long Integer, convert the integer to a real and try again. That way, every command that accepted numeric input now accepted the new Long Integers, too (think of IF THEN ELSE END, WHILE REPEAT etc). The trouble is, they put this test *before* the Automatic List Processing Test, instead of after. Automatic List Processing is not recursive, there's a flag controlling whether you're already in ALP mode or not, so they could just as easily have added the Integer-Real conversion test *after* the ALP test, fixing the above problem. It's what I do in my routines. I think some problem remained even with my routines, but I don't remember any more what it was.. You can try it with Mul in my lib: it works as expected { 1 2 } 3 Mul -> { 3 6 } 3 {1 2} Mul -> idem { 2 3 } {3 4} Mul -> {6 12} > > DIGITS = 1000, pi precomputed > Sqrt(pi) 28 sec > pi - (sqrt(pi)^2) ~ 1E-998 57.7 sec > Gjermund, there seems to be some margin for improvement here. When I calculate Pi to 1000 digits, and then perform << DUP Sqrt DUP Mul Sub >> it takes 28 seconds (<13 for the sqrt), and the result is 0. (the largest error to be expected here is 1.e-999, because both the calculation of Pi, the sqrt, mul and sub guarantee max accuracy.) I always wanted to obtain DIGITS precision. If you specified DIGITS to be 100., then requested VPI, it would return the value of PI rouned to 100. DIGITS. If you set DIGITS to 12., it would return the exact same values as with the built-in reals. That's probably why I never got round to implementing the trig functions ;-)) But Pi calculation (the basis for the trigs) is quite fast. Werner ==== <44ec85ff.0209300653.451a5fbb@posting.google.com>... > > Gjermund, there seems to be some margin for improvement here. > When I calculate Pi to 1000 digits, and then perform > << DUP Sqrt DUP Mul Sub > it takes 28 seconds (<13 for the sqrt), and the result is 0. > (the largest error to be expected here is 1.e-999, because both the I partly disagree ;-) With PI correct rounded to 1000 digits the four last digits are 0199 taking square root with 1000 digits, last 4 digits is: 0669 squaring this number, last 4 digits, correctly rounded, is: 0198 thus error must be 0199-0198 = 1 in last digit = 1 E -999 , which is the number actually calculated (Sorry, I did a mistake when manually converting the above number, which is displayed as 100000....E-1998 ). Gjermund > calculation of Pi, the sqrt, mul and sub guarantee max accuracy.) > I always wanted to obtain DIGITS precision. If you specified DIGITS > to be 100., then requested VPI, it would return the value of PI rouned > to 100. DIGITS. > If you set DIGITS to 12., it would return the exact same values as > with the > built-in reals. That's probably why I never got round to implementing > the > trig functions ;-)) But Pi calculation (the basis for the trigs) is > quite fast. > > Werner > ==== > Gjermund, there seems to be some margin for improvement here. > When I calculate Pi to 1000 digits, and then perform > << DUP Sqrt DUP Mul Sub > it takes 28 seconds (<13 for the sqrt), and the result is 0. > (the largest error to be expected here is 1.e-999, because both the > > I partly disagree ;-) > With PI correct rounded to 1000 digits the four last digits are 0199 > taking square root with 1000 digits, last 4 digits is: 0669 > squaring this number, last 4 digits, correctly rounded, is: 0198 > thus error must be 0199-0198 = 1 in last digit = 1 E -999 > , which is the number actually calculated > (Sorry, I did a mistake when manually converting the above number, which is > displayed as 100000....E-1998 ). > I fully agree! Don't know where the 0 came from.. when I do it now, it shows 1.e-999 > 1) if these big numbers really are relevant > 2) modify the basic multiplication routine, perhaps one can use some 500 > digits multiplications instead of one 1000 digits multiplication > 3) perhaps select another algorithm, perhaps calculate the 1/sqrt(x) > instead, since I too know this is sigificantly faster when digits > about > 200 I use simple Newton-Rhapson iteration.. but doubling the DIGITS at each step, so it is the last step (1000. digits accuracy) that takes the longest. Each iteration consists of 4 adds and 1 divide, so the overall time is in the order of .. a single 1000-digit division. (actually, even less.. for a 1000-digit division divides a 2000-digit number by a 1000-digit number to arrive at an accuracy of 1000 digits, where the last step in the NR-iteration divides a 1500-digit number by a 500-digit number) It's a bit strange.. 1000-digit mul and div take about 17_sec, but the sqrt needs only 14_sec. This means the mul and div can be further optimized;-) But your timings on 100-digit numbers are impressive.. 3 times faster than mine? How do you do it? > > perhaps the end result will be merging with your library > > I will not even try to compete with your PI calculation, presently Im using > some AGM iteration whenever necessary, but I think the trigs are ok ;-) > Feel free to use it. It has taken me quite a while to select the 'best' method, and to implement it in a fast way. Werner ==== <44ec85ff.0210010623.406c572@posting.google.com>... > Gjermund, there seems to be some margin for improvement here. > When I calculate Pi to 1000 digits, and then perform > << DUP Sqrt DUP Mul Sub > it takes 28 seconds (<13 for the sqrt). I have done some improvement, Pi (1000 digits) << DUP FSQRT DUP FMUL FSUB >> now takes 26.5 seconds, <12 for square root Pi(100 digits) square root now takes 0.46 seconds ;-) Gjermund ==== YESHH !! Maybe you will co-operate even more to make the best LongFloat for the 49G: WHIGS-library (I=Integrated :) PS: Do you both support the separation of US from England? > needs only 14_sec. This means the mul and div can be further optimized;-) > But your timings on 100-digit numbers are impressive.. 3 times faster than > mine? How do you do it? > perhaps the end result will be merging with your library I will not even try to compete with your PI calculation, presently Im using > some AGM iteration whenever necessary, but I think the trigs are ok ;-) > Feel free to use it. It has taken me quite a while to select the 'best' > method, and to implement it in a fast way. Werner ==== <44ec85ff.0210010623.406c572@posting.google.com>... > Gjermund, there seems to be some margin for improvement here. > I use simple Newton-Rhapson iteration.. but doubling the DIGITS at each step, > so it is the last step (1000. digits accuracy) that takes the longest. > Each iteration consists of 4 adds and 1 divide, so the overall time is > in the order of .. a single 1000-digit division. > (actually, even less.. for a 1000-digit division divides a 2000-digit number > by a 1000-digit number to arrive at an accuracy of 1000 digits, where the > last step in the NR-iteration divides a 1500-digit number by a 500-digit number) > It's a bit strange.. 1000-digit mul and div take about 17_sec, but the sqrt > needs only 14_sec. This means the mul and div can be further optimized;-) > But your timings on 100-digit numbers are impressive.. 3 times faster than > mine? How do you do it? > In order to get X = sqrt(Y*10^m), X with eg.n=1000 digits let y= digits of Y and add 1000 zeros + 1 if m is odd, thus y is essencially a 2000 digits integer. I also use Newton-Rhapson iteration with doubling of digits, but for some reason (*) I formulated it as x' = x + ( y - x^2) /(2x), thus at final step to get x of 1000 digits I divide a 2000 digit number by a 1000 digit number ;-( However, what may make the difference is: Since the N-R approx. doubles the accuracy of x' for each step, I choose starting digits such that the 2nd last iteration will have much more than half correct digits, then in last iteration, with full precision, I'm guaranteed that I will have the precision required without checking. ( starting digits are either 2*3 or 2*4 ), and except final iteration x is only half of the digits compared to x' x and y I don't need exact result for floating point, but in some cases when I need it for ZSQRT2 there may be an additional iteration. (*) the x' = x + ( y - x^2) /(2x) may in some cases makes it easy to know whether you should add or subtract 1 in last digit in order to get z*2 <= y < (z+1)^2 After your comments I will need to reconsider change to either x' = x+(y/x-x)/2 or x'= (x+y/x)/2 which should be faster for approx. results. Especially since ln and all complex trig functions use AGM iteration with square roots ;-) With large numbers, somewhere between 1000-2000 digits, I suspect that the ^QMul, ^QDiv might changes algorithms, since the running time then increases rapidly. -Gjermund PS sqrt(pi) with 200 digits 1.5 sec, 400 digits 4.5 sec ==== Hey! You could work together and exchange ideas. Veli-Pekka > <44ec85ff.0209300653.451a5fbb@posting.google.com>... X > perhaps the end result will be merging with your library I will not even try to compete with your PI calculation, presently Im using > some AGM iteration whenever necessary, but I think the trigs are ok ;-) X ==== I've some experience in softkeys menus in RPL What I want now is get from the ROM of my 48GX the program that creates the stat menu 96, for modifying it, and assign the new menu to the key LS 5. The standard assignament of this key is: :: MenuMaker ROMPTR A9 60 DoMenuKey ; What must I do for recovering the original program menu from the ROM? ==== > I've some experience in softkeys menus in RPL > What I want now is get from the ROM of my 48GX the program that creates the > stat menu 96, for modifying it, and assign the new menu to the key LS 5. > The standard assignament of this key is: > :: MenuMaker ROMPTR A9 60 DoMenuKey > What must I do for recovering the original program menu from the ROM? this is just the question I discussed some days ago in my post The Menu System. Obviously, one has to get the menu to the stack for manipulatiing it. This just does the command LMN (List Menu Names) from OT48. I put this library on my site below for you, but without any documentation (not yet ready :-). Perhaps you should read in OT49.txt what this powerful command does. The 48-LMN command was programmed completely different from its 49-LMN version because the menu system of these two calculators differ considerably. Menu 96 is a 1-page menu of statistical directory names. These names itself aren't rompointers since dir-names are neither in HARDROM nor in libraries. If you run 96 LMN, you get the list of rompointers { XLIB 168 101 ... XLIB 168 106 }. This is just the result of splitting rompointer XLIB 169 96 which is what is commonly called an array of XLIB-names. Unfortunately, even JAZZ would not help here because JAZZ does not treat arrays of XLIB-names :-) But with LMN you get this list of XLIB names which isn't a correct name here - it's a list of rompointers each of which is itself a lists of 2 items: the head is the menu-key builder and the second item is the program executed if the menu key is pressed. Anyway, if you've got JAZZ, you can read and modify the rompointers in the above list ad libitum. If not, one can still use the above list to append or to omit some items. For modification one may use the command 3tog from Ot48. It allows to decompose and to rebuilt arbitrary lists and programs composed from SysRPL objects, even if these are unreadable without tools like ED from JAZZ. You can also extract rompointers with the command DTYPE from OT49, but clearly, for reading them in detail one needs very strong decompilation tools like JAZZ. Note that the result of any manipulation *must* be a proper argument for TMENU or maybe CST. Study, for instance, the examples under LMN from OT49.txt. Hope this helps - Wolfgang ==== Wolfgang Rautenberg escribi.97 en el mensaje this is just the question I discussed some days ago > in my post The Menu System. > Menu 96 is a... > ..... > Hope this helps - Wolfgang Well. I think my aim is very difficult with my knowledge of SysRPL and my no-knowledge of OT48... After all, I must teach maths (and live!) ==== R Lion escribi.97 en el mensaje Well. I think my aim is very difficult with my knowledge of SysRPL and my > no-knowledge of OT48... Spying the Spectrum's ROM was easier... ==== Beta in my GX, and I have a couple of questions. First, the Erable documentation refers several times to a Beginners' Menu that should appear when the PRG key is pressed. I always get the standard 48 PRG menu from that key, regardless of whether it's in USER mode or not. However, the Erable main menu *does* come up when I press MTH. Second, without Erable installed, pressing Right-Shift MEMORY brings up the MK Filer, but after installing Erable, this key sequence reverts to the regular 48 Memory display. Can anyone tell me why either (or both) of these things are happening, and if there's a way to fix them? -- Wayne Brown | When your tail's in a crack, you improvise fwbrown@bellsouth.net | if you're good enough. Otherwise you give | your pelt to the trapper. e^(i*pi) = -1 -- Euler | -- John Myers Myers, Silverlock ==== >Beta in my GX, and I have a couple of questions. First, the Erable >documentation refers several times to a Beginners' Menu that should >appear when the PRG key is pressed. I always get the standard 48 PRG >menu from that key, regardless of whether it's in USER mode or not. >However, the Erable main menu *does* come up when I press MTH. Second, >without Erable installed, pressing Right-Shift MEMORY brings up the MK >Filer, but after installing Erable, this key sequence reverts to the >regular 48 Memory display. Wayne, My Erable 3.119 works as described. I had to use the beginners menu today to get the original math menu because I did not remember the new short cut for that. My flags are RCLF #8000010045014FF1h #10EC01h. Perhaps it was removed from the absolute version. I am pretty new at this system also. Bill alternate E-dress wtstorey@ieee.org.no.spam.please (Use the obvious) ==== > Wayne, > My Erable 3.119 works as described. I had to use the beginners menu today to > get the original math menu because I did not remember the new short cut for > that. My flags are RCLF #8000010045014FF1h #10EC01h. > Perhaps it was removed from the absolute version. I am pretty new at this > system also. In 3.201 you can get to the original math menu with alpha-rightshift-MTH. I assume it works the same way in 3.119, but I haven't tried it. -- Wayne Brown | When your tail's in a crack, you improvise fwbrown@bellsouth.net | if you're good enough. Otherwise you give | your pelt to the trapper. e^(i*pi) = -1 -- Euler | -- John Myers Myers, Silverlock ==== >In 3.201 you can get to the original math menu with alpha-rightshift-MTH. >I assume it works the same way in 3.119, but I haven't tried it. > I knew it was a simple key combination. I just forgot what and didn't have time to look. Using the other menu gets me there in a few keystrokes. version from a covered port? Bill alternate E-dress wtstorey@ieee.org.no.spam.please (Use the obvious) ==== Bill Storey escribi.97 en el mensaje > My Erable 3.119 works... I had problems with this version (I remember several TTRM's in class using REF with some matrices) so I'm using the 3.2 for years... ==== Wayne Brown escribi.97 en el mensaje > Beta in my GX, and I have a couple of questions. First, the Erable > documentation refers several times to a Beginners' Menu that should > appear when the PRG key is pressed. I always get the standard 48 PRG > menu from that key, regardless of whether it's in USER mode or not. > However, the Erable main menu *does* come up when I press MTH. Second, > without Erable installed, pressing Right-Shift MEMORY brings up the MK > Filer, but after installing Erable, this key sequence reverts to the > regular 48 Memory display. > I never used the Erable assignments, so can't help you with the begginers' menu, and I never had the Filer problem... If this occurs in USR mode, clear this assignment. I think it's MUCH better you work with the new commands looking for them in the main Erable menu and with time, you will see which of them you frequently use and then you will be able of make your own assignments and menus... ==== > I never used the Erable assignments, so can't help you with the begginers' > menu, and I never had the Filer problem... If this occurs in USR mode, clear > this assignment. > I think it's MUCH better you work with the new commands looking for them in > the main Erable menu and with time, you will see which of them you > frequently use and then you will be able of make your own assignments and > menus... how to do the things the documentation says are on it. It's just that I'm one of those people who always want to make things work the way they're documented, whether I need them or not... As for the Filer issue, it turns out that it's not caused by Erable at all, but by the modified eqstk that's installed by the simplified installation procedure after it installs Erable. If I disable the modified eqstk, then RIGHT-SHIFT MEMORY brings up the Filer both in and out of USER mode; but with eqstk enabled it brings up the normal 48 OBJECTS IN { HOME } display. So I've just made a key assignment for Filer on [RS] MEMORY which works in USER mode even with eqstk enabled. RAM cards from Klotz. I've just ordered a 2 MB card from them. -- Wayne Brown | When your tail's in a crack, you improvise fwbrown@bellsouth.net | if you're good enough. Otherwise you give | your pelt to the trapper. e^(i*pi) = -1 -- Euler | -- John Myers Myers, Silverlock ==== Wayne Brown escribi.97 en el mensaje > ...So I've just made a key assignment for > Filer on [RS] MEMORY which works in USER mode even with eqstk > enabled. Ok... but I think that eqstk is for showing equations in pretty print, isn't as I never used it, I can not be sure). About the RAM card you ordered, I hope it works as well as the mine (1MB) ==== > Wayne Brown escribi.97 en el mensaje >> ...So I've just made a key assignment for >> Filer on [RS] MEMORY which works in USER mode even with eqstk >> enabled. > Ok... but I think that eqstk is for showing equations in pretty print, isn't > as I never used it, I can not be sure). I think you're right. It was installed automatically when I installed Erable, but doesn't seem to be required. I think I'll reinstall with the Personalized instead of the Simplified install so I can choose exactly which components to install. -- Wayne Brown | When your tail's in a crack, you improvise fwbrown@bellsouth.net | if you're good enough. Otherwise you give | your pelt to the trapper. e^(i*pi) = -1 -- Euler | -- John Myers Myers, Silverlock ==== while version 12.2002 of Filer1/2 for the 49 was What is new? Now one can easily do ARCHIVE or RESTORE to or from an arbitrary port with a choose box for these commands on a key which is active only if the filer scans a port. Hence, one may keep 3 different backups of the HOME dir (which includes key assignments, alarms etc) in the 3 ports, all under the same port-name HDir. Equal names do not matter since one can RESTORE only from a definite port. This command errors if there is no HDir in the scanned port. Clearly, somebody who often risks a crash can use his favourite expanded filer after a crash only if it is stored in port 2, say, not only key-assigned. The Ar/Rs choose box was checked thoroughly, it is 99% error-proof. I would like to know how 49-users are pleased with the new filer. And I would also appreciate if people working simultaneously with the 48 and 49 would compare the best 48-filer with Filer1 or Filer2. The size-aspect should also count, I need 1.5 KB only :-) Have fun - Wolfgang ftp://ftp.math.fu-berlin.de/pub/usr/raut/HP49/tools ==== I've replaced the standard filer with a key assignment to Filer2 and a command for the port2 Ar$Rs in my custom menu. My only suggestion would be to create a little program to automate the keyassignment and perhaps getting the very usefull Ar$Rs into the APPS menu. I'd really like to make use of the Application menu extention but haven't been able or had time to get my head around how it works. I'd like create something that will do something with the variable selected depending on it's type eg program-run it, Equation-start solver routine otherwise leave on stack etc. As for how good it is? I have a directory on my PC for usefull 49 downloads, it has a special subdirectory dedicated to Wolfgang Utilities. All of them are small, quick and generally usefull. Stephen.N while version 12.2002 of Filer1/2 for the 49 was What is new? Now one can easily do ARCHIVE or > RESTORE to or from an arbitrary port with a > choose box for these commands on a key which is > active only if the filer scans a port. Hence, one > may keep 3 different backups of the HOME dir > (which includes key assignments, alarms etc) in > the 3 ports, all under the same port-name HDir. > Equal names do not matter since one can RESTORE > only from a definite port. This command errors if > there is no HDir in the scanned port. Clearly, somebody who often risks a crash can use > his favourite expanded filer after a crash only if > it is stored in port 2, say, not only key-assigned. The Ar/Rs choose box was checked thoroughly, it is > 99% error-proof. I would like to know how 49-users > are pleased with the new filer. And I would also > appreciate if people working simultaneously with > the 48 and 49 would compare the best 48-filer with > Filer1 or Filer2. The size-aspect should also count, > I need 1.5 KB only :-) Have fun - Wolfgang > ftp://ftp.math.fu-berlin.de/pub/usr/raut/HP49/tools ==== Anyone got them? I'm too lazy to write one from scratch :( -- Computers are not intelligent. They only think they are. ==== Here is the information I got from the HPCC group about their conference and the HP Reps. Oops, maybe you should have been included on the try to get onto it (eric@brouhaha.com). HP intend to continue in the calculator business! But the new people plan to create _only_ designs. The actual implementation and manufacture will _all_ be subcontracted to large Chinese calculator makers. This seems to be a compromise between those senior HP managers who think that still being in the calculator business is an anachronism for HP, and those who think it is a traditional HP product line and should not be abandoned. Two new business models made this way are in the works and should be released about Christmas or New Year. HP told us the model numbers, but asked us to keep this stuff confidential for now. They also said they would continue with RPN - part of the discussion concerned the suggestion that a new name should be used for RPN - HP Logic System or something like that, since the name RPN is used by the other makers as a way to frighten potential customers. Many of us at the conference felt that this is a bit better than nothing, but that the traditional users who depend on HP quality are likely being abandoned :-( Hope this helps! :-) Wlodek -- Richard Garner rgarner@vidnet.net ==== Business models? At last? Can you at least tell us if they will be serious, professional machines, real RPN replacements for the 19BII or ugly ducklings .88 la 39/40? Please? I'm asking because I'm learning accounting and would like a professional machine. Should I get the 17/19BII or wait? -- Thierry Morissette thm47@msn.com > Here is the information I got from the HPCC group about their conference and > the HP Reps. > Oops, maybe you should have been included on the > try to get onto it (eric@brouhaha.com). HP intend to continue in the calculator business! > But the new people plan to create _only_ designs. > The actual implementation and manufacture will > _all_ be subcontracted to large Chinese calculator > makers. This seems to be a compromise between those > senior HP managers who think that still being in the > calculator business is an anachronism for HP, and > those who think it is a traditional HP product line > and should not be abandoned. Two new business models made this way are in the works > and should be released about Christmas or New Year. HP > told us the model numbers, but asked us to keep this > stuff confidential for now. They also said they would continue with RPN - part of > the discussion concerned the suggestion that a new name > should be used for RPN - HP Logic System or something > like that, since the name RPN is used by the other > makers as a way to frighten potential customers. Many of us at the conference felt that this is a bit > better than nothing, but that the traditional users who > depend on HP quality are likely being abandoned :-( Hope this helps! :-) > Wlodek > Richard Garner > rgarner@vidnet.net ==== > Business models? At last? > > Can you at least tell us if they will be serious, professional machines, > real RPN replacements for the 19BII or ugly ducklings .88 la 39/40? > > Please? > > I'm asking because I'm learning accounting and would like a professional > machine. Should I get the 17/19BII or wait? Finally, it would depend on your needs and the time you expect to fullfill them. Can you wait a few months for a new model ? I don't know if the unofficial announcement of a new financial calculator is true or if it would be better than the actual ones (10BII, 12C, 17BII and 19BII). Am I missing anyone ? Many here work around more heavy technical issues, and prefer to program their HP 4X flexible horse-powered machines with their financial needs. I actually own a 49G and bought a 12C, a year ago. In your case, and if I have to choose today, I'd rather buy the HP 17BII model. Hope I'd been useful ==== In message , Thierry >Can you at least tell us if they will be serious, professional >machines, real RPN replacements for the 19BII or ugly ducklings .88 la >39/40? We can't say because they didn't tell us. -- Bruce Horrocks Hampshire England bh@granby.demon.co.uk ==== Rats! > In message , Thierry >Can you at least tell us if they will be serious, professional >machines, real RPN replacements for the 19BII or ugly ducklings .88 la >39/40? We can't say because they didn't tell us. -- > Bruce Horrocks > Hampshire > England > bh@granby.demon.co.uk ==== > Business models? At last? Can you at least tell us if they will be serious, professional machines, > real RPN replacements for the 19BII or ugly ducklings .88 la 39/40? Please? I'm asking because I'm learning accounting and would like a professional > machine. Should I get the 17/19BII or wait? You did not ask from me, but personally I would buy the 19BII It's both AOL and RPN and the solver is fun to use plus there is relatively lot of room for many formulas. Also it's packed with reprogrammed functions. I wonder if HP could do a more feature-rich calculator... Veli-Pekka ==== This might be a newbie question, but what the heck. I used to do a lot of SysRPL development on the 48GX and most of the development I did was done on the PC with IDE programs like HP-IDE. I recently got a 49G and wanted to start doing SysRPL on it, but can't find any IDE's for the PC to do my development on. I have read that the 49G comes with an editor on the calc itself, which is great, but wanted to do my development on the PC. Can you all give me some suggestions to what I should use to do HP49G SysRPL development on the PC? Links to the programs you suggest would be great also. Again, sorry if the question is completely off-base. Doug ==== > This might be a newbie question, but what the heck. I used to do a lot of > SysRPL development on the 48GX and most of the development I did was done on > the PC with IDE programs like HP-IDE. I recently got a 49G and wanted to > start doing SysRPL on it, but can't find any IDE's for the PC to do my HP had a development system called the SDK or Debug2. For some months now, I have been developing version 2 of that system called Debug4x. Debug4x fixes a ton of bugs and adds a lot of usability features to the old debug2. Debug4x, for example, supports both the 48 and 49. It can pop-up a list of RPL verbs if you type the first few characters; it can show stack diagrams for the RPL under the cursor; it can develop input forms using a GUI interface. I used HP-IDE for a long time. Believe me, you would never want to go back! download the program and add you to the beta test list. The download is about 5 MB and comes with a partially completed help system and demos for libraries, directories and simple verbs. -- - - - - - - - - - - - - - - - - Bill Graves RKBA! bgraves@ix.netcom.com ==== the development tools for the 49G named 'Debug2' can be found on www.hpcalc.org There may also another, improved & debugged version named 'Debug4x' be coming soon. However, a progrmmable editor like UltraEdit or (preferred) TSE, along with the 32bit HP tools and Emu48 will do the job as good as some IDE. Raymond Douglas Rohm schrieb im Newsbeitrag > This might be a newbie question, but what the heck. I used to do a lot of > SysRPL development on the 48GX and most of the development I did was done on > the PC with IDE programs like HP-IDE. I recently got a 49G and wanted to > start doing SysRPL on it, but can't find any IDE's for the PC to do my > development on. I have read that the 49G comes with an editor on the calc > itself, which is great, but wanted to do my development on the PC. Can you > all give me some suggestions to what I should use to do HP49G SysRPL > development on the PC? Links to the programs you suggest would be great > also. Again, sorry if the question is completely off-base. Doug ==== I have two unused copies of the HP48G Owners Manual. One is still shrinkwrapped. They are $25 or best offer. ==== Can anyone tell me if there is any difference between the HP49 commercialized in France and that in English speaking countries? Clearly the packaging and booklet are in French, but is there any difference in the calculator itself (as there was for the HP48 versions)? Hwyl Paul ==== > > Can anyone tell me if there is any difference between the HP49 > commercialized in France and that in English speaking countries? > Clearly the packaging and booklet are in French, but is there any > difference in the calculator itself (as there was for the HP48 > versions)? Replying to myself... I checked the FAQ, which made it fairly clear. Today I took the plunge, and bought what was probably the last 49G in a shop in Grenoble. Sadly, the 48<->49 cable is missing so I might have to take it back :-( (it wouldn't be so bad if I didn't have a 48GX). Which leads me to... does anyone know where I can get my hands on a PC cable (since of course the 48 serial cable won't work). A bientot Paul ==== > > Which leads me to... does anyone know where I can get my hands on a PC > cable (since of course the 48 serial cable won't work). > > A bientot > Paul I made my own HP48 cable using a tiny 4-pin connector that's sold in electronics stores as a CD-ROM Audio Cable. When I got my HP49, I noticed that the four pins at the lower left are in the same order as the 48, and the little connector fits over them without touching any of the others. The rest of the cable, the PC end, came from a mouse. That means that it's nice and flexible. Since you can buy surplus mice for the equivalent of $5, and the CD-ROM cable for about $1, you get a PC cable for either the 48 or the 49 for about $6. Bill ==== I live in France and in early 2002 I finally decided to buy a HP49G (being a proud and happy owner of a HP42S since 1990). Surprise: I had a lot of difficulties to find one! Then I read on the WWW discussions on HP closing down their calculator line. In my panic I called many dealers in the Paris region and finally when I found one I bought it right away, in an original package, but obviously (lots of dust on the box) a VERY old machine, serial ID01400681 (made in Indonesia), VERSION yields HP49-C, Revision #1.16, Copyright HP 1999. So I thought I had bought one of the last available (but one of the earliest) HP49G in France. Note: This is probably only possible with the french dealers who don't care and don't know about HP machines (I won't tell the name here but is is a big supermarket chain selling lots of electronics). Anyway. They promised me to provide a serial PC connection cable but still I haven't got it, I think they just don't all you guys who keep the world going with this kind of free information exchange. Johannes ==== Please can others try the following boolean expression with Bool49, Truth Table and OneTerm (A AND C) OR (A AND D) OR (B AND C) OR (B AND D) Truth table gives me a table of all zeros, and OneTerm returns [A B C D] correct list of variable and an empty list {} incorrect. I expect it to return {5 6 7 9 10 11 13 14 15}. Do others see the same behaviour or is it something specific to me? ==== > Truth table gives me a table of all zeros, I get the correct truth table. > and OneTerm returns [A B C > D] correct list of variable and an empty list {} incorrect. I expect > it to return {5 6 7 9 10 11 13 14 15}. I get {5 6 7 9 10 11 13 14 15}. Do you have anything stored in the variables A, B, C or D? If not, provide your flag list, and I'll see what I can find. ==== > Truth table gives me a table of all zeros, > > I get the correct truth table. > > and OneTerm returns [A B C > D] correct list of variable and an empty list {} incorrect. I expect > it to return {5 6 7 9 10 11 13 14 15}. > > I get {5 6 7 9 10 11 13 14 15}. Do you have anything stored in the variables > A, B, C or D? I don't think I have A,B,C or D defined. I have tried typing A [ENTER] in the HOME directory and the directory where I was orgininally using Bool49v2.0, this I'll call my Working Directory. In both cases I end up with 'A' on the stack. Similaryl with B,C & D. Typing VARS to give a list of varriables in the working directory, and intermediate directory between Working and HOME and the HOME directory gives me a list of variables, A,B, C or D are not present in any of these lists. Conclusion I don't think I have any variables called A,B, C or D. Have I missed any possibilities here? Just for completeness I tried (DOG AND HORSE) OR (DOG AND ZEBRA) OR (FISH AND HORSE) OR (FISH AND ZEBRA) and again got the all zeros truth table and null result from one term. > > If not, provide your flag list, and I'll see what I can find. I presume by the flag list you mean the system flags accessed via [MODE][FLAGS] The following have a check mark 19,31,36,40,41,42,44,57,68,72,73,82,90,114,117,119,128 > > > > Just in case it is of significance ROM version 1.19-6 MArk ==== > Have I missed any possibilities here? No, you were very thorough - I would only have checked with 'A' ENTER etc. > I presume by the flag list you mean the system flags accessed via > [MODE][FLAGS] Yes, you can recall your flags with the command RCLF. Such a recalled list can then be stored again with STOF - that's a fast way to change back to a previous set of settings. > The following have a check mark 19,31,36,40,41,42,44,57,68,72,73,82,90,114,117,119,128 You like the small font, eh? ;-) > Just in case it is of significance ROM version 1.19-6 I bet you have the internal word size set to 3 or less - run the command RCWS and see if it returns 3 or less? Now do 64 STWS, and use Bool49 again :-) ==== > Have I missed any possibilities here? > > No, you were very thorough - I would only have checked with 'A' ENTER etc. > > I presume by the flag list you mean the system flags accessed via > [MODE][FLAGS] > > Yes, you can recall your flags with the command RCLF. Such a recalled list > can then be stored again with STOF - that's a fast way to change back to a > previous set of settings. > > The following have a check mark 19,31,36,40,41,42,44,57,68,72,73,82,90,114,117,119,128 > > You like the small font, eh? ;-) > { # 12B8A41050FF0h #180000000000000h #8052000002020188H #400h} > Just in case it is of significance ROM version 1.19-6 > > I bet you have the internal word size set to 3 or less - run the command > RCWS and see if it returns 3 or less? Now do 64 STWS, and use Bool49 again > :-) OK you hit the jackpot. I have a program that executed 3 STWS. The program takes the output of OneTerm and converts the result back to a boolean expression just to make reading easier. I am trying to work out if a boolean expression or list is the best presentation format. The boolean expression has the advatage of being able to use BView. Need to work on the program a more following this discussion. > > MArk H ==== Do RCLF VP X > I presume by the flag list you mean the system flags accessed via > [MODE][FLAGS] The following have a check mark 19,31,36,40,41,42,44,57,68,72,73,82,90,114,117,119,128 ==== I just uploaded the second part of the update to the sequences, series and limits marathon to hpcalc. The marathon contains now in addition to the already existing contents, how recurrence sequences can be handled with the HP49G and programs for finding generating functions and converting (some kinds of) recurrences to their analytic closed forms, so that the programs then can find bounds condensation points etc. If someone can't wait until the next update of hpcalc, don't hesitate Greetings, Nick. ==== I've used my hp49g for a couple of years now and I'm completly saticfied. But today it started doing a thing that it never done before, it asked me Change to approx mod? or something like that, and if I didn't do it, the result couldn't be calculated! I've never tried that before. I always just keep it in exact mode. What flags is missing now? Another thing that started today, I can't write 85^4 (or similar expression) without my calculator calculates the value right away! And it doesn't write it nicely! It simply puts out the value without the EE. What's wrong here? Last thing, is just a questions about the hp49g in general. How come the Ti89 is faster at calculating integrals? Eg. x*exp(-x^2) integrated from 0 to infinite. The Ti89 gives the answer right away but the hp49g needs a lot of time to think about it, before it gives the answer 1/2. And the integral x^2*exp(-x^2) integrated from 0 to infinite can't even be calculated on my hp49g, but it can be done on the Ti89. Anyone with a reason for this? Anders Misfeldt ==== > But today it started doing a thing that it never done before, > it asked me Change to approx mod? or something like that, That is because you provide numeric input - any number with a decimal point is considered numeric input. If you have used numeric input before without the calc asking you permission to change mode, it's because you have had silent mode ON earlier (set flag -120). > Another thing that started today, I can't write 85^4 (or similar > expression) without my calculator calculates the value right away! And > it doesn't write it nicely! It simply puts out the value without the EE. Huh? Example, please. > Last thing, is just a questions about the hp49g in general. How come the > Ti89 is faster at calculating integrals? Eg. x*exp(-x^2) integrated > from 0 to infinite. The Ti89 gives the answer right away but the hp49g > needs a lot of time to think about it, before it gives the answer 1/2. The TI89 spends 0.16 seconds on this, while the HP49G spends 4.36 seconds. The reason is simple: The TI89 uses heuristics widely, which is a huge advantage when dealing with integrals like yours. Specifically, the TI89 looks in a table if it knows the integral, and plugs in the limits - the HP49G *calculates* each integral, hence it'll always spend an amount of time applying calculus rules. The HP49G could also have used heuristics, had it been implemented that way, but Bernard Parisse chose not to. I guess because of development time and lack of space - the latter is most likely. When the integrals get tougher, the HP49G will sometimes beat the TI89 with a large margin, but whenever you calculate school book examples (a very wide group of problems), the TI89 knows the answers pretty much right away. Some examples, where the TI89 is alot slower than the HP49G: '1/(X^4+1)^4' 'X' RISCH, TI89: 113.27 seconds, HP49G: 9.80 seconds '1/(SIN(X)-2)^3' 'X' RISCH, TI89: 36.74 seconds, HP49G: 12.57 seconds 'EXP(X)*COS(X)^4*X^3' 'X' RISCH, TI89: 42.51 seconds, HP49G: 15.58 seconds > And the integral x^2*exp(-x^2) integrated from 0 to infinite can't even > be calculated on my hp49g, but it can be done on the Ti89. There does not exist a closed form solution to the integral of 'X^2*EXP(-(X^2))' - that is, not without utilizing special functions like ERF(X). Such special functions aren't supported by either of the two calcs, and even the TI89 calculates this integral numerically. So can the HP49G, but it won't do that automatically, like the TI89 - you have to use ->NUM or be in numeric mode. With 5 decimals accuracy, the TI89 spends 10.50 seconds on the above, while the HP49G spends 17.08 seconds when ->NUM is used - they agree on the result ;-) There exist a number of expressions the TI89 can't integrate, where the HP49G can (as well as the other way around, of course). For example: '(a*LN(a)-2*X^2)/(a*LN(a)^2)'|a=X^2-1 -> 'X/LN(X^2-1)' '(2*X^2+1)*EXP(X^2)' -> 'X*EXP(X^2)' '1/(1+SQRT(X^2-1))' -> ...alot... ...alot... -> 'LN(X)^2/(EXP(X)+1)+LN(EXP(X)+1)' There exist also factorizations that the TI89 can't do, while the HP49G can. There are even a larger number of limits the TI fails at, while its solver is very good. The two machines have very different powers. ==== >>Another thing that started today, I can't write 85^4 (or similar >>expression) without my calculator calculates the value right away! And >>it doesn't write it nicely! It simply puts out the value without the EE. Huh? Example, please. > Ok eg. I want to calculate 4^84, so In RPN-mode I write: 2: 4 1: 84 then hit the Y^x buttom and then I get this on my display: 1: 374144419156711147060143317175368453031918731001856 instead of just: 1: 4^84 or 1: 3.741444E50 maybe with more or less decimals. How can that be? Anders ==== >>Another thing that started today, I can't write 85^4 (or similar >>expression) without my calculator calculates the value right away! And >>it doesn't write it nicely! It simply puts out the value without the EE. > Huh? Example, please. Ok eg. I want to calculate 4^84, so > In RPN-mode I write: > 2: 4 > 1: 84 > then hit the Y^x buttom and then I get this on my display: > 1: 374144419156711147060143317175368453031918731001856 Right! > instead of just: > 1: 4^84 Use '4^84' > or > 1: 3.741444E50 Use 4. 84. [Y^x] > maybe with more or less decimals. > How can that be? You have several choises Use the one that you want/need/prefer. > Anders > Veli-Pekka PS: You may also try the ALG-side or simply use `back-quotas` for direct ALG VP ==== > How come the Ti89 is faster at calculating integrals? Eg. x*exp(-x^2) > integrated from 0 to infinite. The Ti89 gives the answer right away but the > hp49g needs a lot of time to think about it, before it gives the answer 1/2. Maybe the TI-89 evaluates the corresponding indefinite integral and subtitutes the endpoints (taking limits if needed)? Since its indefinite integration is very fast, that would explain it. > And the integral x^2*exp(-x^2) integrated from 0 to infinite can't even > be calculated on my hp49g, but it can be done on the Ti89. Anyone with a > reason for this? Not really answering the question but posing a related one: does the HP49G automatically try for an approximate solution if it can't find an exact one? Also, is there an equivalent for the approx() function on the HP49G? -- Bhuvanesh ==== > > How come the Ti89 is faster at calculating integrals? Eg. x*exp(-x^2) > integrated from 0 to infinite. The Ti89 gives the answer right away but the > hp49g needs a lot of time to think about it, before it gives the answer 1/2. > > Maybe the TI-89 evaluates the corresponding indefinite integral and > subtitutes the endpoints (taking limits if needed)? Since its > indefinite integration is very fast, that would explain it. > > And the integral x^2*exp(-x^2) integrated from 0 to infinite can't even > be calculated on my hp49g, but it can be done on the Ti89. Anyone with a > reason for this? > > Not really answering the question but posing a related one: does the > HP49G automatically try for an approximate solution if it can't find > an exact one? Only for polynomials, if you have numeric factorization activated. For other things you must use ROOT or related commands. >Also, is there an equivalent for the approx() function > on the HP49G? Yes, ->NUM and XNUM. Greetings, Nick. ==== >Also, is there an equivalent for the approx() function > on the HP49G? > > Yes, ->NUM and XNUM. Those don't sound like equivalents. Remember that approx() not only numericalizes exact numbers but also temporarily uses APPROX mode in calculations (that are between the parentheses). Bhuvanesh. ==== > Those don't sound like equivalents. Remember that approx() not only > numericalizes exact numbers but also temporarily uses APPROX mode in > calculations (that are between the parentheses). So does ->NUM and XNUM. One of the differences between the two is that XNUM leaves the calc in numeric mode. Real mode: Sigma(X=-5,2000,X) EVAL (exact mode): 2.08 seconds Sigma(X=-5,2000,X) EVAL (numeric mode): 14.23 seconds Sigma(X=-5,2000,X) ->NUM (exact mode): 14.51 seconds Sigma(X=-5,2000,X) XNUM (exact mode): 14.75 seconds Hence, ->NUM and XNUM runs in approx mode. ==== > Sigma(X=-5,2000,X) EVAL (exact mode): 2.08 seconds The TI89 spends 3.81 seconds here, by the way... > Sigma(X=-5,2000,X) EVAL (numeric mode): 14.23 seconds ...but only 7.75 seconds here ;-) ==== I am trying to test for a number being <0 and >0 (well two separate numbers really) I have used the %0< command but I do not know how to deal with this, if I run the program on the calculator all I get is an external. Is this system rpl speak for TRUE? or am I way off course. An example would be most helpful. MC Reply-To: peter@dutw1479.wbmt.tudelft.nl ==== > > I am trying to test for a number being <0 and >0 (well two separate numbers > really) I have used the %0< command but I do not know how to deal with this, > if I run the program on the calculator all I get is an external. Is this > system rpl speak for TRUE? or am I way off course. An example would be most > helpful. > MC The 48/49 will sow external for many pointers that are not USERRPL commands. In your case it may very well be the TRUE pointer. A way to find out is to decompile it with ->S2 an example it checks if the number is greater than 0 and does different things to it depending on the outcome :: CK1&Dispatch 1 :: DUP %0> ITE (if then else) :: %8 %+ (executed if true) ; :: xSIN xSQ (executed if false) ; ; ; @ -- ir. P.F.Geelhoed Delft University of Technology Laboratory for Aero & Hydrodynamics Leeghwaterstraat 21, 2628 CA Delft, The Netherlands +31-15-2786656 / +31-15-2782947 (fax) peter@dutw1479.wbmt.tudelft.nl ==== Or set flag -85, then the stack will be displayed with the SysRPL decompiler, and TRUE will display as TRUE (if extable is present), just as with SSTK in Jazz (at least the HP48 Jazz version). Set flag -92 before compiling SysRPL. ((o)) . . ==== Everytime I goto graph, i get invalid ppar and cannot graph a damn thing. Is there a way to reset the ppar to default? I tried the ppar values from the hp48gx manual, but still get an error ( I have a 49 and no info in those manuals) Any help would be greatly appreciated. Brandt ==== > Is there a way to reset the ppar to default? It will be automatically recreated with defaults if first purged from the current directory. > I tried the ppar values from the hp48gx manual The default 'PPAR' list is: { (-6.5,-3.1) (6.5,3.2) X 0. (0.,0.) FUNCTION Y } If you copied from a 48gx manual and typed 0 as the resolution instead of 0. and if your 49 was also in Exact input mode at the time, then this would cause Invalid PPAR! The RES command will safely convert even integer type to real, however. . ==== Put 'ppar' on the stack and then purge it. Everytime I goto graph, i get invalid ppar and cannot graph a damn > thing. Is there a way to reset the ppar to default? I tried the ppar > values from the hp48gx manual, but still get an error ( I have a 49 > and no info in those manuals) Any help would be greatly appreciated. > Brandt ==== >Put 'ppar' on the stack and then purge it. > Don't forget you may need to do this for other PPARs in other directories above the current one if you are in a subdirectory when you start, until you find the one(s) that is invalid. Dennis ==== >Put 'ppar' on the stack and then purge it. > Don't forget you may need to do this for other PPARs in other directories above > the current one if you are in a subdirectory when you start, until you find > the one(s) that is invalid. Dennis ==== Seen: > Don't forget you may need to do this for other PPARs > in other directories above the current one, > if you are in a subdirectory when you start, > until you find the invalid one(s). Only the current directory is searched for 'PPAR'; if not already present in the current directory, a new 'PPAR' will be created there if needed, regardless of any others existing in higher directories (any exception would be a bug, so note it if you can find it). Some brand new 49 coding tended to forget the here part of various SysRPL words, causing several problems, but this should not have happened. . ==== was very angry. What do you think? Please, understand me. I still do love HP for those golden years. It might make no sense to you, but I'm sharing it only here to be aware of how can a loyal customer feel. These were the planned headers in the beginning. point? I was right. Right you are. Certainly Carly Fiorina should know the trained skills her employees have. So YOU ARE familiar with the newsgroup I referenced... and you still had to search for key pressure. (Thought in that case you'd already filled forms regarding users requirements, inquiries... Oh Lord, I must be so naive !!!) And the only reference that YOU could find to the differences (mentioned), between the CN and the ID calculators (in the more than 96,000 threads in comp.sys.hp48), ... concluded Key pressure on both calcs are about the same. 3 Newtons, to be specific. (???) Later on about 3 N of pressure to depress the key where the older HP Calculators required 1.3 to 1.7 N. N being Newtons, in case you were wondering. See http://www.hpcalc.org/details.php?id=3023 for details. I attached the thread, so you can read it again. I hope if you'd forgive me saying so but when you write (or speak) to an engineering student, you should probably know that he may well know the meaning of [N]ewton, probabbly better than the acronyms CN and ID (standing for Chinese and Indonesian) you'd used assuming everyone knows them that way... although It's not hard to figure them out. According to the URL you'd told me about, I've been there. Regarding 49 vs 48, by Per Str.9amgren from Karlstad, Sweden. Per.Stromgren@ein.ericsson.se dated on 1999/10/16 14:35:18, it is another Proof that the 49G keyboard is twice as hard to press as the 48 keyboard, possibly leading to a repetitive stress injury !!! __________________________________________________ http://www.hpcalc.org/details.php?id=3023 (...) In order to base things on facts I set out to measure the downforce needed to press a key on the 48 vs the 49 (plus a 32 thrown in for comparison). This is the result: HP48SX, 9 years old, heavy use 1.4 N(*) HP32SII, 4 years old, light use 1.7 N(*) HP49G, brand new 3 N(*) So it's obvious that the 49 _is_ harder to use in a sense! ... I also tried to measure the throw of each key and found that the old rocker type keys (32 and 48 in this case) had the force knee (key click point) at about some 10-20% of the full throw, whereas the 49 keys can be pressed some 50 percent before hitting the knee (no pun intended). This also makes up the feeling of the keys: the new keyboard can be pressed without going into action, the old did not do that. I have found that the keyclick function is useful on the 49; I suppose its there because of the changed keyboard. Question to HP insighters: does the above findings reflect the design implied by user requirements or is this a just a cheaper solution than the old keyboard? Per Str.9amgren, (*)The figures are accurate to about +/- 15%. Method: put the calc on a ordinary 0-1kg kitchen scale, read the weight, press a key and read the weight when the key clicks. Subtract the two numbers and convert to force i Newtons. (...) __________________________________________________ If you need further assistance, ... ??? You must be joking ! You proved both my points, the fact of the key pressure difference needed for ea model and that your excellent new hp web site development don't meet its goal of providing accurate information, along with the fact that is the best web reference together with http://www.hpcalc.org/ and that you can be rude enough to answer a former loyal customer the way you did. Tks for nothing, Carlos Can't believe how HP is dealing with her customers. You're so lucky to have a job today. You're doing everything right not to deserve it. ----- Original Message ----- I AM familiar with the newsgroup you reference. I did search for key pressure. The only reference that I could find to the differances between the CN and the ID calculators concluded Key pressure on both calcs are about the same. 3 Newtons, to be specific. I am also aware that the HP 49G requires about 3 N of pressure to depress the key where the older HP Calculators required 1.3 to 1.7 N. N being Newtons, in case you were wondering. See http://www.hpcalc.org/details.php?id=3023 for details. If you need further assistance, please reply to this message. You may also find technical and troubleshooting information along with software updates, on our HP Customer Care Web site at: http://www.hp.com/ Kellie Customer Solutions support agent for the new HP * Our advice is strictly limited to the question(s) asked and is based on the information provided to us. Problems and solutions may depend on the nature of your system environment and various other parameters that are unknown to HP; therefore, HP cannot assume any responsibility or liability. Please be advised that technical information changes as new data becomes available, therefore, HP recommends that you regularly check our Customer Care Web site for possible updates at: http://www.hp.com/go/support HP shall not be liable for any direct, indirect, special, incidental or consequential damages in connection with the use of this information. Original Message Follows: ------------------------- Dear Amanda, If you'd join http://groups.google.com/groups?hl=en &group=comp.sys.hp48 and read older postings (as many from the support department should), you'd probably notice the users do notice the difference on pressure needed for its keys (among other user's impressions), between both models, as well as I did. I couldn't find much help (technical and troubleshooting) information, nor recent software updates, on your HP Customer Care Web site at: http://www.hp.com/ _____ -----Original Message----- You mentioned it crashes frequently. Is there rhyme or reason in it crashing? Meaning does it happen when you do anything in particular or just random. If you need further assistance, please reply to this message. You may also find technical and troubleshooting information along with software updates, on our HP Customer Care Web site at: http://www.hp.com/ Amanda Customer Solutions support agent for the new HP ... _____ Original Message Follows: ------------------------- ... Although this service is provided as a means of technical support for our customers, we appreciate any feedback about our company's products and services. proper individuals. We value your comments. I'm forwarding your question about your HP Calculator to the appropriate group. If your ... Alex Customer Solutions support agent for the new HP ... _____ Original Message Follows: ------------------------- Dear Jean, ... As a HP freak, I've been using HP calculators for more than 15 years now. A year ago a friend of mine brought me from Atlanta, Georgia the latest state-of-the-art HP 49G. Although I disliked its design, it is a great machine. Even though it crashes frequently, it still is a good equipment. I hope the management won't close the Calculators department ACO (along the Handhelds area), as I'd heard. I believe the Indonesian version of the 49G I've got, has a much harder pressure needed for its keys than the Chinese recent release, which I also regret haven't bought instead. I used to complain years before, of the quality of argentine dealers (linked with HP Argentina's lack of support). I'm afraid that the whole HP experience faded away along with its superb quality. An (almost) cheap device, even if its branded HP, will still be a CHEAP device. I'm pretty angry with the fact that my trust in HP was betrayed. I don't complain on the recent merge, I've only want the HP quality back, the one that seems gone for good, since the end of the 80's. You may copy this message to Carly Fiorina. In fact, I would like you to do it so. Buenos Aires, Argentina ==== I don't find anything wrong with the 3N keyforce but the required keytravel of almost 50% before the positive registration of a hit is unbearable! Earlier JYA said that HP has considered the 10B II keyboard for the 49G (being internally almost the same) If this is true, then this is what HP should do: Replace the HP 49G keyboard with the HP 10B II and call it HP 49G II The better screen cover, new HP 1.19-7 ROM released as ROM 2.00 and new User's Guide (basically the Urroz Books for the 49G Vol I and II proof-red and updated) and a 5MHz CPU speed to gain 20% more speed plus to enhance the 15360 baud serial to 19200 would do the trick !! People would buy this model (at least I would :) and with good User's Guide even new users would finally get a grip on it's capabilities, and the keyboard which has the click on it, would certainly please people. Veli-Pekka