A lot of disturbing compiler complaints (char vs. unsigned char) about detox HOT 5 CLOSED

Hey, Doug, you're my hero of the day!

Thank you so much. Your approach to stick with signed chars and deal with the fallout is far superior to anything else.

from detox.

Thanks. I've noticed similar warnings from Clang (on Travis) and CLion. I'll address this in an upcoming release.

Unfortunately, most of the standard C functions take signed chars as their inputs, and the character transliteration logic in detox needs unsigned chars to work properly.

The code works with the implicit casts; I'll probably go with explicit casts in the v1 series and introduce more breaking changes in a major release.

from detox.

Yeah, this is big, and potentially breaking. I'll address it in v2.0.0.

from detox.

I feel with you.

My C brain is rusty at best, but maybe a set of wrapper macros/functions for the standard c functions with the correct casts in place would be the shortest path to success.

from detox.

Thanks @frispete ! I looked at your suggestion of creating wrappers around the standard C library, too, but ended up going with this approach when I realized it was only one of the filters that was failing, and I could fix it easily.

I also incorporated some of the gcc flags I saw from your output, adding protections against overflows and overruns. I like those a lot.

Thank you very much for the push in the right direction!

from detox.