Yeah, I think I'm just going to use it for tasks that my calculator isn't capable of on its own, like solving for x.

1. Some sources say that all addins are limited to 2MB, and some sources say that the limit is only on the CG10 and CG20, with the CG50 having a much higher limit. Are you trying to remain compatible with the CG10/20, or does the CG50 actually not have the higher limit?

2/3. I get it now.

Thanks for fixing AC/On. I think Casio should have designed their operating system so if an application doesn't handle that key, it will be closed or, even better, give me an "Application not responding" dialog box to allow me to either continue or stop. Unfortunately, I doubt that they listen to feedback from the users.

I don't believe that addins can be larger than 2M on the CG50, at least the French equivalent (Graph 90+e) does not.
For interruptions, it's not that easy to implement, especially with ressource constraints. On PC for example, I can compile Xcas with C++ exceptions enabled, and this make error handling much much easier, but that would not work for KhiCAS, because it requires more memory and there is this 2M limit.

That 2MB limit sounds really annoying. Anyways, good job fitting all of this functionality in 2MB.

How do i enter the cube root of a number. For example, 8^(1/3) is 2 but khicas returns the same fraction.

8^(1/3)=>*

You can get the => by pressing the arrow that is normally used for assigning variables.

There is no auto-simplification in KhiCAS, except for fractions of integers. You must explicitly call a rewrite command, like simplify from F1 menu or a command shortcut like
=>+ (partfrac)
=>* (factor or ifactor)
...

Me again. I've been using KhiCAS for real, and it is great. Some of my friends are wondering how I managed to get a CAS on a calculator that cost less than their TI-84s. Anyways, I think I figured out the real cause of the slow closing issue. The corrupted binary I had earlier certainly didn't help, but it wasn't the cause. And I'm not sure if I mentioned it, but the issue did come back immediately after the first time I used it. Here's what I've found:

Closing the app and opening something else will take unacceptably long, unless I clear everything before I do so. This suggested to me that it had something to do with saving the session. Manually saving from the file menu also takes about as long as closing the app and opening another one. It still seems to autosave even if I manually save.

Unfortunately, I could not find an option to disable auto saving, so I have to remember to clear the screen every time I want to switch out of KhiCAS. Is there a way that I just haven't found, or do you have to recompile the app because of size limitations?

I cleaned out my calculator, optimized the storage and tried again. This time, even without clearing the screen, it did not take an unreasonable amount of time to close the app and manually saving also takes a lot less time.

I contacted Casio about some issues. Their excuse for the graphing app being slow is because they "place emphasis on the drawing process" whatever that means, and the reasoning for not having a default interrupt handler, along with other poor design decisions, is "proprietary information." I suspect that whatever caused the flash storage to be so slow until I optimized it is a result of Casio's poor decisions. So the really slow app closing issue almost certainly is not your fault.

Yes, leaving KhiCAS auto-saves session.xw for next opening. This can take some time the first time you save, but should take less time later unless session.xw is larger. And this takes indeed more time if you flash storage area has many files and not much room left.

Hi,

I defined a function x^3+2x^2+4x-5:
first by
P(x):= x^3+2x^2+4x-5 using shift DEL to get :=
and then by
x^3+2x^2+4x-5 -> P(x)
The first way of defining the function yields an answer like x**3+2*x**2+4*x-5; so quite straight forward.
The second version yields an answer like
// Parsing P
//Success // compiling P
lambda xSad x**3+2*x**2+4*x)-5
Question: what happens here?
Are := and -> different in defining functions but the same in defining variables?

Hi another thing for graphs:

I am positivly surprised about the graphs in khicas on the Casio for zooming and scrolling.

I remember for having read something that XCas is also used on the HP Prime. But there the CAS Graphs are somehow difficult to use: So far I remember it is possible to get them on one screen but no easy zooming and scrolling. Or did I overlook something on the HP Prime?

Best regards
Andreas

I can not reproduce, both return a function (displayed as x->x**3+2*x**2+4x-5 in Xcas mode or lambda x**3+2*x**2+4x-5 in Python mode). Maybe you did P:=x**3+..., defining an expression in P (instead of a function).

You're right. The UI on the Prime does not support graphs from commandline as well as KhiCAS, because HP expects users to graph with one of the Prime App, not from a CAS command.

KhiCAS is now available on the Casio Fx-9750GIII
https://www-fourier.univ-grenoble-alpes.fr/~parisse/casio/khicasen.g1a
The documentation is the same as on the Prizm
(https://www-fourier.univ-grenoble-alpes.fr/~parisse/casio/khicasioen.html)

That’s impressive - will it work on fx-9860gii SD (I have the "USB Power Graphic" not "USB Power Grapihic 2" model of it - so should be able to have upto 2GB SD card on it but all the cards I have are llarger SDHC type - so need to order and older version of sd card to testunless you know that Khicas won't run on my model then there is no point of buying SD cart to test)? Do you use separate sdks and code separately g1a and g3a? Or something like mini sdk which I‘ve heard supports both somehow?

The answer is unfortunately most certainly no.
I'm using two SDK (they have a common part and separate libraries).

Thank you for confirming - would it be possible to reduce the g1a file size to below 1,572,864 bytes somehow to test on fx-9860GII please?

He's already fought somewhat to make a functionally reduced version of KhiCAS fit into 2 MB, so cutting a further quarter of that would require even more work... assuming the goal can be reached in the first place.

Thank you - I remember 2mb was a struggle. I don’t recall if you compile on windows or linux? Perhaps there is a way to split the addin into two? On prizm this would allow increasing functionality further. The first addin could launch the second and vice versa. Probably difficult but should achieve the goal of full functionality. For g1a I don’t know if it is even possible to launch any addins from within another addin and given the overall storage memory is so small on 9860gii, only SD models could potentially swap addins between storage and sd memory but that’s even more difficult but hopefully doable (at least with manual swapping of addins) - all depends if possible to split Khicas into two addins. Thank you for whatever you can do and let me know if i can help experimenting with it if it is compilable in windows? By the way I think Simon Lothar had mini sdk which supported system calls for both prizm and fx9860 - you may find him on http://www.casiopeia.net/ - maybe it will help with not having to use different sdks. Also he may share some ways of swapping addins in runtime and perhaps he knows if at all possible to have larger than 2mb addins for prizm... I doubt it’s possible

I don't think you can reduce the size of KhiCAS to 1.5M. Probably you can save 100K by reducing the UI to a small CAS shell, and another 100K by removing further some CAS commands, but then there is a large common kernel you can't really reduce.
I compile on a linux box (debian 9).
I don't plan to work on a splitted version of KhiCAS addin, I have no idea how to cross-link the two parts and I don't think it's worth the effort. It seems the 2M limit is somewhat artifical, Casio could probably raise this limit with a few hours work.

Makes sense - thank you