Fix compile warnings in RiceVideo

[GoogleCodeExporter]

I noticed the CFLAGS var in the makefiles doesn't include -Wall. I tried
compiling RiceVideo with -Wall and got a lot of warnings. This issue is to
track cleaning up warnings.

Original issue reported on code.google.com by ebenbl...@gmail.com on 27 Mar 2008 at 8:15

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Yeah the original RiceVideo makefile didn't have -Wall in it.  When I merged the
projects together I saw all those warnings and removed -Wall for 
RiceVideoLinux.  I
agree it would probably be better if it compiled cleanly with -Wall, but it's 
going
to take a fair amount of work to clean all that junk up.

Original comment by richard...@gmail.com on 28 Mar 2008 at 5:04

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

I fixed all the char* to const char* legacy issues for normal compilation and 
debug. This was the majority of problems. We were redefining a char* datatype 
and 
confusing the compiler. Anyway... will commit these soon. 

Adding -Wall now mainly gives comparison of signed to unsigned, and unused 
variables. I should be able to fix that quickly enough.

Original comment by sknau...@wesleyan.edu on 4 Apr 2008 at 12:29

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Original comment by sknau...@wesleyan.edu on 4 Apr 2008 at 12:30

[GoogleCodeExporter]

Thanks for working on this. Did you add -Wall to all plugins as well, or are 
you just
working on the main mupen64plus code? I know the graphics plugins have quite a 
few
warnings too.

Original comment by ebenbl...@gmail.com on 4 Apr 2008 at 4:45

[GoogleCodeExporter]

I probably should have said I fixed all the char* to const char* legacy issues 
in 
Rice. There were around half a dozen in glN64 (I also fixed but haven't 
committed 
that yet), and loads in Rice. No problem though. :)

To be clear, I fixed all const char* and all other general warnings in Rice in 
commit 97. 

I then added -Wall to Rice and got loads more errors. I'm currently removing 
these 
and got 95% of them gone now. As I said above, most of it is really dumb stuff. 
Whoever added some parts didn't realize that sizeof() returns size_t or keep 
track 
of whether the value they were comparing was int or u32int or that arrays in 
structs and classes need {}. 

Once I finish up and do somemore testing, these fixes will be my next commit. 
For 
the life of me, though, I can't figure out why this is this warning in 
TextureFilters.cpp. In our software additive blending code:

BYTE* pdst...
*pdst++ = *(pdst-1);

It gives an operation not always defined. I guess since *(pdst-1) <=> pdst[-1] 
and 
pdst is a pointer it could be a NULL pointer. Is that what the compiler is 
complaining about?

I decided to move on and fix all the simple errors first. 

I don't get any general warnings in any other parts of mupen64plus. Glide64 
compiles with -Wall and I don't get anything there. If we want to cull warnings 
in 
glN64 and main I don't mind doing it, but that's another issue.

I think that even after I commit the next batch of changes to Rice we should 
either 
adjust this tracker or close it and start another one to address the style of 
the 
code. Rice is very clearly written by many different people in many different 
styles. Some simple cleanup would make finding bugs and adding features easier. 
My 
commits seemed extra long since I did a little style cleanup on functions that 
gave 
warnings as I went.

Original comment by sknau...@wesleyan.edu on 4 Apr 2008 at 5:24

[GoogleCodeExporter]

Original comment by sknau...@wesleyan.edu on 4 Apr 2008 at 6:33

• Changed state: Started

[GoogleCodeExporter]

Great work. Changing this issue to be specific to RiceVideo.

Original comment by ebenbl...@gmail.com on 4 Apr 2008 at 8:12

• Changed title: Fix compile warnings in RiceVideo
• Added labels: Component-RiceVideo

[GoogleCodeExporter]

[deleted comment]

[GoogleCodeExporter]

Thanks, commit 98 gets rid of everything but the pointer issue above. Someone 
else 
might want a look at that so we can close this Issue. :)

Commit 99 and 100 fixes all non-debug warnings in glN64. The Debugging system 
for 
that plugin seems not to exist (adding -DDEBUG generates errors, and big ones 
like 
missing typedefs and files, not just warnings) so this should be discussed 
before 
anything is done. I started Issue 40 to deal with that.

Original comment by sknau...@wesleyan.edu on 4 Apr 2008 at 9:00

[GoogleCodeExporter]

With the removal of tabs, Richard fixed the last warning, hence the issue is 
now 
fixed! Probably another issue should be opened for general code style cleanup.

Original comment by sknau...@wesleyan.edu on 8 Apr 2008 at 11:23

• Changed state: Fixed