A639
===
Subject: Re: Has HP fixed the keyboards in newer models?
It seems that a new calc may be ok but since there has not been any
official statement by hp on a new design ,if there really is in fact any,
and how the new design is devoid of any reliability problems or multi-bounce

problems what anyone person says is just speculation. There is no guarantee

that any new design will continue to be produced anyway.
He believes that it is a newer type Is it the newest type?
There is no mention of keyboard reliability problems on the TI new group.
===
Subject: Re: More help wanted at OpenRPN.org
My understanding is that support for other filesystems can be added
within eCos. I will look into this more closely for you, though and
post my findings.
===
Subject: Re: HP 39G+ Problem to transfer Software from PC
I can't remember whether the 39g+ has a (Wire) option on its SEND/RECV
menu (I don't have one here). If it does then that may work. It pops up
a file selection window on the PC that lets you download via Xmodem.
===
Subject: Re: HP 39G+ Problem to transfer Software from PC
PS. A lot of the games that were written for the 39G & 40G will not work
on the 39g+. Some use system calls that no longer work on the new chip.
Others are simply so fast on the new chip that you can't play them. I
recall one which had a frog hopping across the road avoiding cars. On
the 39g+ the cars were unavoidable because they were travelling at the
speed of light.
===
Subject: Re: HP 39G+ Problem to transfer Software from PC
That is surprising, almost no entry points have moved between the two
machines.
None of the supported did either.
JY
===
Subject: Re: StdIOPAR sysRPL?
TW
===
Subject: Re: Learning SysRPL
Right, but I usually use the LongFloat
I wonder who is the genius behind it?
It would be great to have LongFloat Matrices in ARM-C
The whole library could be conveted (slowly, but surely)
to C language and thus have both more digits and great speed
Even zapping temporarely to 120MHz would be nice for LongFloat
===
Subject: Re: Learning SysRPL
The interface to the hp49g+ is mine. All the rest which are doing the actual

work is not mine, it has been made by C. Bond . I found it on the web.
The interface is basically like this: The user-rpl program expands the
matrices on the stack, issues a command number, then runs the hpgcc
program.
The hpgcc program pops this command, and decide which subprogram to run (
thats the first switch statement, similar to case statement on the

hp
calcs.
The subprogram allocates memory, if out of memory the subprogram is aborted,

otherwise it pops (reads) the matrix elements from the hpstack.
Then the actual c program is called to do the work.
The subprogram pushes the result and any messages on the stack and
de-allocates any memory.
A few questions I'd appreciate any advice:
1, Is it necesssary to explicit de-allocate memory, or will the hpgcc system

do this ?
2, I usually don't want leave messages in the middle of the stack, but
prefer to put it upon exiting the program. I know I can write these messages

to a buffer using sprint, but it seems to be clumsy, I have to modify the
calls to all subsroutines, and I have to keep track of both the buffer and
the position in this buffer.
Is there any way to write to the stream stderr, how do I set this up in
order to later get this back and write it to the stack?
I learned the basics of ANSI C based on the examples included with the hpgcc

and some tutorials found on the net. I do however recommend to get a proper

text book. That will save you a lot of time.
I don't know the difference between c and c++, that was not required to
learn to program this calc ;) , but do a search , I'm sure there are
comparisions to be found.
Wrt. solving, first of all,
if you solve the system Ax=b by x = INV(A)*b , ie. inverting A, then you
will have to multiply with b afterwards, thus you are doing perhaps twice as

much work as necessary.
Also, all this unnecessary complicated number crunching will make the
results less accurate.
Now, on the hp calculators, for square A, it is better to treat the matrices

similar as numbers, simply put b and A on the stack and hit the divide
button.
With regard to how these solvers actually work, start with a search for LU
solvers or LU decompostion. This idea / algorithm was on the top 10 list of

algorithms of the century. You will not waste your time if you study this.
Gjermund
===
Subject: Re: OpenRPN Production Plans
X
Will it accommodate 48gII, 39g+, 40gs
and possible same-sized future models
???
===
Subject: Re: HP48G Vs HP49G (speed concern)
That's fissio, Qonos would be fusion bomb!
===
Subject: Euro 2006 in Finland summer 2006
Suggestions offcially opened
1) Cyrille de Brebission @ HPQ! - Your presense is needed!!
You could announce the HP-58GX II++ Turbo Lazer in Finland
Perhaps in June/July?
===
Subject: Re: Ok: HP is Joking!!!
As a matter of fact, i have rom 2.06 installed and i get much relief by
I used my 49g+ all day long without any problem.
--
____________________________
Marco Polo