Fails to compile with GCC 5.0: narrowing conversion in koepac_voice_archive.cc

[jbeich]

-Wno-narrowing doesn't help to avoid the following error found via DPorts. Maybe use int16_t instead.

src/systems/base/koepac_voice_archive.cc:142:27: error: narrowing conversion of '255
' from 'int' to 'int8_t {aka signed char}' inside { }
     0x7e, 0x81, 0x7f, 0x80};
                           ^
src/systems/base/koepac_voice_archive.cc:142:27: error: narrowing conversion of '25$
' from 'int' to 'int8_t {aka signed char}' inside { }
src/systems/base/koepac_voice_archive.cc:142:27: error: narrowing conversion of '25$
' from 'int' to 'int8_t {aka signed char}' inside { }
...

[minirop]

if it does not go over 255, uint8_t shouldn't be prefered?

[eglaysher]

uint8_t probably should be prefered; given some of the libraries that rlvm links in, I wouldn't expect rlvm to compile with signed characters. (Most of the code which assumes characters are always unsigned predates rlvm by years. Like most of the image and audio parsing code, the koepac decoding code originally comes from xclannad.)

I suspect a switch to uint8_t...might work here. But now I'm wondering exactly why the variable d (line 213) is signed. That might be significant... Every time I've touched some of these internals, I've made a hash of it...

[minirop]

I look at it, and I don't understand either since d is only used as an array index and cast to an unsigned char each time.