Re: svn commit: rev 5638 - in trunk: . subversion/bindings/com subversion/clients/cmdline subversion/libsvn_client subversion/libsvn_delta subversion/libsvn_diff subversion/libsvn_fs subversion/libsvn_ra subversion/libsvn_ra_dav subversion/libsvn_ra_local subversion/libsvn_ra_svn subversion/libsvn_repos subversion/libsvn_subr subversion/libsvn_wc subversion/mod_dav_svn subversion/svnadmin subversion/svnlook subversion/svnserve subversion/svnversion subversion/tests subversion/tests/libsvn_delta subversion/tests/libsvn_diff subversion/tests/libsvn_fs subversion/tests/libsvn_repos subversion/tests/libsvn_subr subversion/tests/libsvn_wc

On Wed, Apr 16, 2003 at 10:48:40PM +0200, Branko ??ibej wrote:
>...
> Greg, what absolutely astouds me is that you probably didn't compile and
> test on Windows at all with the generated scripts before you removed the
> handwritten ones. Sorry, but that's totally against our policy to run
> the tests before non-trivial commits.

We make all kinds of changes without testing on all platforms. Most
developers don't have access to all of the platforms we support.

But I'm not sure that's a good excuse for me :-), as the change was
specifically regarding one platform. And yes, I trusted the feedback rather
than actually running the tests.

How about this: I have copies of Visual C 6.0 and Visual C .Net. I'm going
out of town for a week starting Friday, but I'll get VC6 installed at home
and give it a best effort. There is also a good chunk that I can do right
now, without actual compilation on Windows (working on that now).

> So now we're almost back to where we were before I started hacking the
> Win32 port -- i.e., we don't have a working Win32 port. Somebody with a
> Windows box will have to spend a lot of time getting the buld up to
> speed again.

You listed the following items:

    * mod_dav_svn doesn't build
    * most of the test programs are missing
    * apr-iconv is missing
    * the .exe files don't contain version info
    * win-tests.py needs to be taught about the new build location

I just checked in a change for the apr-iconv stuff. That would seem to get
us back to "the port works except for mod_dav_svn". We're then missing the
tests and version info.

I should have a bunch of the changes for this checked in today, and then
I'll personally test it later tonite.

> That somebody won't be me. I simply don't have the time these days; if I
> had, I'd have tweaked the generator long ago to DTRT (yes, that would
> have taken me a lot longer than just maintaining the old scripts did).
> I'd intended to maintain the old scripts so that we (I?) had more time
> to get the generated ones working to my satisfaction.

Fair enough. I missed your earlier post, broke the build, and will take
responsibility to get it working again (as much as possible before leaving
for a week).

> As far as I'm concerned, it's now down to three choices: a) dig into
> gen-make.py immediately to get it working in a few days, b) reinstate
> the old scripts, or c) live without Win32 for at least the next couple
> of releases. As I said, I don't have time for a).

(c) isn't an option. I'm working on (a). Can we see where we end up by
tomorrow?

Thanks,
-g

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org