|Re: Printing Scientific Calculators - there were some non-HPs...|
Message #9 Posted by Bill Wiese on 23 Aug 2005, 12:07 a.m.,
in response to message #8 by Howard Owen
Yes, it's more costly. But remember you are starting out with the shell of an HP-97. Those aren't cheap, so we are definitely talking a high-end system here, before we even start with the computing platform.
I'm only building this because I have a 97 or two hanging around, one prob broken, can't recall.
Since you are going to switch emulations, that implies some sort of OS, even if it's just DIP switch settings.
Why would you think that? Not at all.
You're making the situation way too complex. Your whole baseline system without the application or emulation layer is 100X more complex than the calc emulation and original application!
On power-up, say, one could hold down  and  keys for HP21,  and  for HP25, etc. (Or any kinda variation on a similar theme.)
This could simply latch a CPU GPIO pin in turn connected to encoder to select, say, an 8K bank of EPROM containing emulation code. The keyscan software, external to emulated firmware, would monitor for an "emulation switch" and change contexts. Or, just loads a few pointers to new code body and opcode decode/opcode emulation routines. Either way, this is maybe 50 bytes of code or a dozen lines of C code.
My definition of OS includes task switches, semaphores/mutexes, etc. even for a tiny 2-4KB RTOS. None of this is not needed here. We have one thread of operation w/occasional interrupt (if even necessary).
I know that Eric's software runs on embedded Linux because I've ported it to pdaXrom on a clamshell Zaurus. So if I were doing the work, Linux would save me time and effort.
I have not seen the code lately but the core of Eric's emulation would be at least nearly platform independent. It's just moving data around in some emulated registers and performing some very very simple operations upon them. It wouldn't be hard to recast that to entirely neutral platform if necessary. The Linuxy stuff is only for display and KB traps and probably quite separable.
But what really is attractive to me about a fully modern platform housed in a well loved high-end antique case is the possibility of the wizardry being invisible on the one hand,
and very obvious on the other. What I mean is that an HP-97 emulation running on this platform could be nearly indistinguishable from a "real" 97.
So would mine. And that's what Eric is probably talking about too - doing a 41C-like project on a 97 platform.
[quoteA low end system could achieve that too, I concede your point. But the high-end system could serve out the same calculator software over an integrated network to my desktop. [/quote]
Why go to that grief? Just run the regular Ericware emulation. No need for 'server'.
Eric has some really, really cool software in Nonpareil. It all works with an X11 interface. I have no doubt that he could adapt most of it to run with physical keys instead of virtual ones, but the graphical whiz-bang could come along for free!
It is very cool - for a PC Win or Linux box. On raw native simple hardware it's not needed, and I'd bet he'd agree with me.
So what I see is a calculation appliance that could be the best of all possible worlds.
Ooooh, statements like that scare me ;)
would have a desktop PC interface as well as the physical one.
I'd think at best just squirting text files over RS232 or onto a CF or SD card would be fine.
It would have enough computational power to make it a serious contender for the kinds of problems that are solved on PCs nowadays. (Matlab, R and many other such apps run fine on Linux, of course.)
The only reason calcs had higher-order stuff like that was because calcs came of age when minicomputer time was valuable. As calcs evolved, PCs were too expensive for late HS and also colleget math students.
Anyone doing 'big math' on a calc has my sympathies. For that I use Matlab/Maple, etc.
Calcs are good for general 'playing with numbers'. Don't even need Solve & Integrate (unless it were for core of a calc-based application that I wanted - for example, ballistics for me.)
The HP48, 49 got so complex, to me, that they're lost causes. They solve problems that can be more appropriately solved elsewhere. I had a 48SX and gave it away. Someone else gave me another 48 and I gave that one away too.
San Jose, CA USA