What am I doing wrong? about ctod

[dkorpel]

Your first issue reduces to:

#ifndef MAX_MATERIAL_MAPS
    #define MAX_MATERIAL_MAPS 12
#endif

unsigned int x;

And the problem lies with the hacky header guard remover like I suspected. It assumes all code to translate is between the #define and #endif and leaves the rest of the file untranslated. Working on a fix right now.

[schveiguy]

OK, the preprocessor directives seem to really mess with ctod. If I comment out all #ifdefs and #endif and #define, then it seems to do a LOT better. I can work with this.

[dkorpel]

It probably thinks that your define is a header guard and removes it.

Edit: looking at the source file you linked, the problem is bigger than that

I'll look at the code later.

[schveiguy]

My plan is to comment out the ifdefs anyway, and leave them there for reference, so for now I'm OK with it. It would be good to allow that without having to comment manually though.

One thing jumped out at me immediately -- it does not replace unsigned char with ubyte or unsigned int with uint.

[dkorpel]

One thing jumped out at me immediately -- it does not replace unsigned char with ubyte or unsigned int with uint.

That's weird, I'm explicitly testing unsigned int:

ctod/source/ctod/test.d

Lines 21 to 22 in b33a2f2

	@("types") unittest {
	test("unsigned int x0;", "uint x0;");

I added a test for unsigned char and that passes for me as well.

[schveiguy]

I'll push the C file I translated to a branch, and maybe you can discover why it didn't do that.

[schveiguy]

I think I sniffed it out. Looking at all the cases where unsigned was still present were all cases of sizeof(unsigned int) -> unsigned int.sizeof. So I think maybe that translation is overriding the other?

In any case, here is the file that produces those: https://github.com/schveiguy/draylib/blob/rmodelsctod/raylib/rmodels2.c

[dkorpel]

I fixed type translation inside sizeof(), I'm now looking at your original failure in rmodels.c

[schveiguy]

BTW, I think this is going to help me a ton. Parts that I was just going to leave as C I may be able to translate all of them.

Thanks for doing this, I would never have gotten this far!

[dkorpel]

I'm glad you find it useful too. It really does change the scale at which you can translate C, I would translate a 1000 line single-header library by hand, but would not have translated tens of thousands of lines like in glfw-d without a tool like this.

[dkorpel]

This issue is fixed by be78cc4 , please open a new one when you find something else.

[schveiguy]

raylib's rcore took me about a month of small free time pockets to translate. And that was about 6k LOC.

I'm almost tempted to run ctod on it and see how close it is to my hand-ported version lol.