[PATCH] Fix type inconsistencies in stdint.h

Dave Korn dave.korn.cygwin@googlemail.com
Tue Apr 7 12:38:00 GMT 2009


Corinna Vinschen wrote:

> OTOH, we already had to change int32_t and uint32_t from long to int to
> avoid warnings.  Given that we already changed that anyway, I'm wondering
> if it isn't more sane to align the least and fast types as well.

  Well, if there was ever a time to do it, now would be that time, and I'll
happily go update GCC to accord with whatever we decide to do.  I can't say
what kind of incompatibilities might arise, as it's not an easy thing to
google uses of these types specifically in exported rather than internal APIs.
 It's possible things like codec libraries and heavy graphics number-crunchers
might specify these types in externally-visible definitions but I haven't done
an audit.

  As long as we don't change the size, binaries will still interoperate fine,
except where changed name-mangling prevents linking, but e.g. structs and wire
or file formats will remain unchanged.

  I think on balance, it's probably a reasonable idea, but I haven't done a
detailed analysis of the risk so it's possible I've overlooked something
disastrous.  Since 1.7 is still experimental (albeit stabilising rapidly), I
guess we could even just go ahead, and revert it iff problems arise.

  CGF?  You asked a couple of questions and then dropped out of the thread for
a couple of days.  Have you reached any conclusions?

    cheers,
      DaveK




More information about the Cygwin-patches mailing list