Streetwalrus, is an OS considered a "large project" that OOP would be good for? A lot of people think C++ isn't good for an OS, that C is better.

Shouldn't you quote?

Who? Linus?

My calculator (Nspire CX CAS 3.6) crashes and hard-reboots after I run Khicas. Any ideas?

Just to make sure I didn't mess something up by changing things I re-built the source with no modifications except for changing printf in the example to nspireio's functions, setting up console, etc. It still gave "Not enough memory.".

No mallocs that I can find earlier on...all it does it make a couple of variables, setup the global variables, etc. Nothing big, and no mallocs that I can see.

Oh, and crafti is very nice amazed at how fast it goes.

It's not a breakpoint, it's a return value. The original function:

int
init_mem(void)
{
    if (n_tokens <= 0)
        return false;
    if ((scratch = (token_type *) malloc(((n_tokens * 3) / 2) * sizeof(token_type))) == NULL
        || (tes = (token_type *) malloc(n_tokens * sizeof(token_type))) == NULL
        || (tlhs = (token_type *) malloc(n_tokens * sizeof(token_type))) == NULL
        || (trhs = (token_type *) malloc(n_tokens * sizeof(token_type))) == NULL) {
        return false;
    }
    if (alloc_next_espace() < 0) {    /* make sure there is at least 1 equation space */
        return false;
    }
    clear_all();
    return true;
}

I edited the return values to be 1,  (((n_tokens * 3 )/ 2 ) * sizeof(token_type)), 3, and 0, respectively, and edited the function that init_mem(); returned to so it would print the return values, and continue without error if the value was 0. Therefore I could track where the fail was coming from.

I don't think the "breakpoint" was intentional: it was in the code to begin with. Maybe I'm misunderstanding you.

Instead of putting a breakpoint in malloc I did this:

    if ((scratch = (token_type *) malloc(((n_tokens * 3) / 2) * sizeof(token_type))) == NULL
        || (tes = (token_type *) malloc(n_tokens * sizeof(token_type))) == NULL
        || (tlhs = (token_type *) malloc(n_tokens * sizeof(token_type))) == NULL
        || (trhs = (token_type *) malloc(n_tokens * sizeof(token_type))) == NULL) {
        return (((n_tokens * 3 )/ 2 ) * sizeof(token_type));
    }

So if that block (the one that is failing) fails, I'll know the largest block size it's allocating. The return value is 2400. Is that what you were asking for?

Quadratic77

Yes, I saved the flash as nand.img (maybe I saved it in the wrong location, where should I?). And when I loaded the emulator up again, it said to "press I to load default operating system" or something like that. And it did load the default OS again, just like the first time.

Now, if I change DEFAULT_N_TOKENS to 10 and make the "out of bounds" error to <10, it fails at this:

if (n_tokens <= 0){
        return false; }

(in init_mem() -um, that's supposed to be a ';' and a ')', not a smiley...

So it seems like the DEFAULT_N_TOKENS HAS to be at least 100....and, the documentation says mathomatic can run on a few mb's of RAM, so I shouldn't have to hack it down lower.

What do I want to do with exceptions in C? I don't know, but the makefile that came with mathomatic apparently had it in.

Debug it: I can't. The emulator requires boot1 and boot2 and polydumper only dumps my boot1 for some reason, which I can't figure out.

And please pardon my apparent stupidity on this:

nspire-gcc  -Og -g -Wall -Wshadow -Wno-char-subscripts -Wno-unused-variable -marm -mlong-calls  -fno-exceptions -DLIBRARY -DVERSION=\"`cat ../VERSION`\" -DSILENT=1 -DSECURE -DHANDHELD    -c -o lib.o lib.c
cc1.exe: error: argument to '-O' should be a non-negative integer
make: *** [lib.o] Error 1

-Og throws the error I expected it too...where am I wrong here?

Oh, and if I compile with arm-none-eabi-gcc, the calc freezes on the document screen right when I click on mathomaticndless.tns. I'm curious: why then and not after it loads the console?