[OTR-dev] Reproducible builds of pidgin-otr for Windows

[Ian Goldberg]

On Mon, Mar 21, 2016 at 03:44:48AM +0100, HW42 wrote:
> > OK, built.  The above tarball, whose hash is itself:
> > 
> > 22c6a6c250ca277e36ffed1ff1bd6d366209664471c80c3c507f8811ba5c28da  pidgin-otr-4.0.2-repro.tar.gz
> > 
> > generates these files for me:
> > 
> > 19f315c8317105a89d5b6fde6c2caa63ffa2e150df5fb0d67ee20ac985b0b191  pidgin-otr-4.0.2.exe
> > 7e9dc2175591d7aabc9f96e737817fa917f3e4441b62727bb9730e516c47822e  pidgin-otr-4.0.2.zip
> > 
> > How about yoU?
> 
> I can reproduce this in an Ubuntu Trusty chroot (/proc and /dev/shm need
> to be mounted).
> 
> I tested (most) of the variations from [0]. I found only one problem:
> In Makefile.mingw:148 it should be LC_ALL=C instead of LANG=C else the
> locale settings can change the sort order of the files in the zip.

Ah, good catch, thanks!  Pushed to git.

> > This means it can't find /usr/i586-mingw32msvc/share/aclocal/libgcrypt.m4.
> > Aha: it's probably looking for the *native* one and failing.
> > 
> > But that said, we shouldn't actually have to run autoreconf (or
> > bootstrap) at all.  Can you try just removing "./bootstrap" from
> > INSTALL.mingw and see what happens?
> 
> Another option is to set ACLOCAL_PATH:
> 
>   export ACLOCAL_PATH=/usr/i586-mingw32msvc/share/aclocal

Yes, but we really don't want to be re-autoconfing, anyway.

Thanks,

   - Ian