On Jan 30, 2008 7:51 AM, Christian Stimming <[EMAIL PROTECTED]> wrote:
> > Hi Dscho, > > Quoting Johannes Schindelin <[EMAIL PROTECTED]>: > > first let me thank you! It is awesome how much you accomplished, and I > am > > sure you spent quite some time working on it. > > You're welcome. > > >> The "not so good" news is that I'm still not sure how to organize that > >> build in the ./release.sh form. Instead, I still have a long manual > >> recipe... whatever. > > > > Hey, I pulled back a little, because I was exhausted, and that one mail > > (you know which one I mean) really made me rethink my efforts. > > > > But now I think it is time to integrate (and I'll put you in as author) > > it into release.sh. I can do that. > > Sure. I'll wait to see how this will look like in msysgit.git. > > >> * I compiled libtool-1.5.22: > > > > I'll probably just add /src/libtool/release.sh to fetch && build && > > install it, and then add a check for libtool >= 1.5.22 into > > /src/subversion/release.sh. > > I'm actually unsure whether it has to be this latest version. What I > observed is that in msys/temp the libtool installation was missing > some parts, hence "libtoolize" as called by ./autogen.sh didn't run > successfully. The easiest fix for this here was to install this very > latest version, but it's up to you to check whether an older version > might be sufficient. > > >> * I got a zlib1.dll, moved this to /bin/zlib1.dll, and created the > >> corresponding /lib/libz.dll.a as described in my earlier email; the > existing > >> /lib/libz.a I removed or renamed. > > > > Maybe we should take the msys-z.dll from the recent work/new-ssh branch? > > Could be, yes. I'm also not so sure which files are really needed; the > point is if there is only the libz.a, the linker will error out with > "-lz not found". I don't know whether the libz.dll.a or zlib1.dll was > the important part here to satify the linker. > > > If you don't mind, I'll take it further and probably come back to you > with > > something usable by tomorrow night. > > Yes, that sounds good. > > > You are my hero, > > Hee hee... > > Christian I've also been able to build a git-svn that passes most tests (the symlink/executable bit tests fail). I basically did the same things Christian did plus I had to do some extra tweaks. The biggest difference was for the created dll's, I kept getting an error from perl about not being able to remap to the parent process. Googling said that you needed to run cygwin's rebase on the dll's. Doing this fixed the problem, but obviously something is wrong there. Did you have this issue Christian? I also used a different libtool (from gnuwin32) instead of building the latest to save time. As for building subversion, that was the hardest part. I built zlib, apr, apr-util, and neon separately with minor tweaks to the LDFLAGS so that I wasn't getting static linking. I installed each one as it was built. Because of the cygwin piece and also editing build generated files, I'm not sure this will be easy to convert to one of the release scripts Dscho. I will try to start over and retrace my steps, but I don't know if this is an easy ./release.sh and you're done process. Mike
