Re: [MiNT] Keith S. makes Subversion available on Atari!

[Mark Duckworth <mduckworth@atari-source.com>]

On Sat, 2006-12-09 at 12:17 +0100, Guillaume wrote:
> First many kudos to Keith! :D
> 
> This was requested not so long ago (see
> <http://sparemint.atariforge.net/mailinglist/Mailing-Lists/MiNT-List.200510/402174642.20051006140355@der-ingo.de.text>).
> I wanted it just as badly then, as Subversion had started to phase out CVS
> at work and on interesting open-source projects: that is why I write this
> little cheering-up and "advertising" message, as this was originally only
> announced on dhs.nu (http://bbs.dhs.nu/coding/index.php?request=2233).
> Keith, please of course feel free to comment!
> 
> For the record I once tried to get it to cross-compile, I at least figured
> I did not *required* mmap() (at least for the client part), contrary to
> premature suppositions made
> (<http://sparemint.atariforge.net/mailinglist/Mailing-Lists/MiNT-List.200510/4345402E.2040409@coolrunningconcepts.com.text>).
> But for subversion-1.3.0 and apr-0.9.7 at that time, I had to use the trio
> library, nice it is not needed anymore!
> My own stumbling block IIRC was that it required programs to run during
> build phase, so I was out of luck with cross-compiling, as I had neither
> native hardware nor emulated working MiNT development setup (which I am
> about to achieve only now... BTW anyone interested in an up-to-date
> SpareMiNT-from-scratch procedure?).
> 
> So perhaps AtariForge should provide Subversion hosting now! ;)
> 
> And perhaps some good fellow will make RPMs out of the tarballs?
> 
> Speaking of RPMs where should new ones go? On the original SpareMiNT site?
> Or does http://dev.sparemint.org/ effectively replace the former and is
> now officially live?!? If this is the case I think an "official"
> announcement would be needed!

http://dev.sparemint.org will replace the other one and it will be
official someday, but it's not now.  When it's official it will be
accessible from http://www.sparemint.org.  I know it's been over a year
but there's issues that need fixed and right now I'm working on the IT
department of a huge but new company which is causing 12-16 hour days
for me.  The main issue that I uncovered really is that when someone
uploads a new revision of a package there's no rules for superceding.
Hence, when abc* packages get updated frequently they queue up on the
packages that need approved list.  Which is kind of silly, the next
build should supercede the previously disapproved build....

Frankly, I'm reversing my thinking on the approval system.  It simply
doesn't have enough attention.  I think a subset of wellknown people
will be able to move packages from testing to production repositories
and everyone else with an account will be able to upload right into
testing.  We're small enough that that makes more sense than an approval
system that nobody pays attention to.

Thanks,
Mark