`byte' typedef?

Werner Koch wk at gnupg.org
Tue Jul 2 10:36:02 CEST 2002

On Mon, 1 Jul 2002 15:09:23 -0500, David Champion said:

> But that only works if one of the entities being compared is a void*.

I was more thinking about assigning.

> If one isn't, then you have to prototype as void* (gnupg doesn't) or

memcmp takes void *. strcmp takes char * which is corect in this case.

> Yes, some compilers have such options. But what is the solution to
> this problem (excessive warnings) that you would accept into the code
> base? Would you accept casts? Or prototype changes, where there's no

No.  All type conflicts are to be solved ... except for unsigned char*
and signed char* mismatches; we can't get rid of them and I can't
think of a machine where these are really distinct types like function
and data pointers are.

> particular conflict? Or would you rather I submit a patch setting the
> Solaris CC to "cc -erroff=E_ARG_INCOMPATIBLE_WITH_ARG", and cause gnupg

IBM compiler have ption just for the signed/unsigned mismatch.

> I just want to see warnings in stock code disappear, and I'm trying to
> help make it happen.

I don't see any warnings with my compiler when using -Wall
-Wcast-align -Wshadow -Wstrict-prototypes.



More information about the Gnupg-devel mailing list