Re: Any active maintainers using MinGW/MSYS?

[Eli Zaretskii]

> Date: Mon, 27 Mar 2006 18:03:15 -0500
> From: Earnie Boyd <address@hidden>
> Cc: address@hidden, address@hidden
> 
> Quoting Eli Zaretskii <address@hidden>:
> 
> >> Date: Mon, 27 Mar 2006 08:26:01 -0500
> >> From: Earnie Boyd <address@hidden>
> >> Cc: Eli Zaretskii <address@hidden>, address@hidden
> >>
> >> >  >> 4) You'll need to replace the delivered config.guess and config.sub
> >> >  >> with the ones in /usr/share/libtool; alternatively you can modify
> >> >  >> config.guess and config.sub in the package to add a filter for MSYS
> >> >  >> whereever you find one for CYGWIN.
> >> >
> >> >  ez> I'd suggest to ask Paul to create an msys/ subdirectory and put 
> >> > there
> >> >  ez> the modified versions of config.guess and config.sub.  
> >> > Alternatively,
> >> >  ez> if the modifications can be done by some Sed script, add just that
> >> >  ez> script to the distribution.  The W32 build instructions 
> >> should then be
> >> >  ez> modified accordingly.
> >>
> >> No, that isn't necessary.  My hope would be to use the official version
> >> of make in native win32 (MSVCRT) build from within MSYS.
> >
> > Sorry, I don't understand what you are saying.
> >
> > You said that config.sub and config.guess need to be modified for them
> > to DTRT for the MSYS build of Make, so I suggested to make the
> > necessary magic as part of the Make distro, either as alternate
> > versions of those scripts, or, better yet, to have a script that would
> > edit them as part of the build, before we run the configure script.
> >
> > How is this related to using ``the official version of make in native
> > win32''?  We are talking about running the configure script, not about
> > running Make.  What am I missing?
> >
> 
> I don't want MSYS to become Cygwin.  Jonathan had asked how to build an 
> MSYS version (using the MSYS runtime instead of the MSVCRT runtime) and 
> I explained how to do that.  I want it to be as difficult as possible 
> to build source for MSYS because I intentially want people to not want 
> to use the MSYS runtime for their own purposes.

Does that mean you want us to reject any patches to the sources that
are conditioned on __MSYS__?

(I don't think you can prevent people from building an MSYS port, no
matter how hard you try, and so making it as difficult as possible is
just being harsh to them.  But since you are the MSYS maintainer, you
are free to decide that you don't want to help them.)

> At the same time it is my desire to use as many native (requiring 
> MSVCRT runtime) binaries as possible within the MSYS toolset framework. 

Are there any issues that prevent the native Windows build of Make
from becoming _the_ MSYS Make that is compatible with the rest of the
MSYS toolset?  If so, could you or someone else publish the list of
those issues?

Btw, just so that I understand: what should be the minimal set of
packages in the MSYS toolchest, ideally?  I'm guessing that it's the
shell and possibly the patched config.* scripts, but perhaps there's
something else?  And, with enough patchwork, isn't it possible that
the need for MSYS will disappear, and users will be able to just use
native ports of Bash and other utilities?