[Uml-devel] Project coordination

[Thomas Baumgart]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi all,

on Friday 19 July 2002 11:24, Luis De la Parra wrote: 

> > Hi,
> >
> > Why do you want to block the cvs? I think it is not necessary to block th
> > e
> > cvs. A better way is, to define some TAGS and open some branches for test
>
> ok, I admit I'm not a cvs-expert, but I am not propossing to block cvs
> permanently (that we all have to send the code to some "coodrinator" and
> only he can commit.. NO, I didnt mean that)
>
> I only suggested to block it for a couple of days, to make sure no one
> commits code in one or two days and give time to the coordinator to
> straight things up and fix all problems in cvs. after that cvs should be
> open again and people can commit direct (of course, AFTER updating its own
> copy and making
> sure they dont break anything and that no one else modified the same code
> they
> modified, in which case they should first solve all conflicts before
> commiting)
>
> branches would be a very good idea if, for example we want to stabilize uml
> for the next release, which everyone wants to make soon without stoping
> new-features development. we could have the release branch and the
> development branch, but now I'm only talking about stabilizing things so
> that we can later branch and merge more easily.

The project handbook has it all ;-)

In kmymoney2 we faced the same problem, came up with a solution, documented it 
and ran through it once. Worked great.

http://kmymoney2.sourceforge.net/phb/version-mgmt.html


> everyone can still code on its cvs copy, but then after cvs is fixed, the
> first time you commit have to get a fresh copy, merge your changes to your
> new
> cvs copy and then, if everything works ok,  commit. from that point on its
> easy.. it's only getting everything fixed the first time which is
> problematic since everyone seems to have
> completly, uncompatible, cvs version (for example, the CR/LF problem, etc)
> and just applying diffs is not easy. there has to be a lot of manuall
> merging. (diffing the CR/LF files will tell you they are completly
> different, and in reallity maybe only 2 lines changed and all the others
> can be fixed with dos2unix, but this has to be done and tested manually by
> someone. that's why I inist in first ONE person fixing cvs, and then the
> rest of us has to adapt to this "new" code
>
> so, I still propose to block, fix, and then reopen.
> if Jonathan is willing to do this work, Paul agrees, and Jens and everyone
> else accept this we could take Thomas' and Jens' tips so that everyone can
> commit with out accidentaly fucking things up for everyone.

I agree, the CVS mess has to be cleaned up first. During this time, no commits 
should be performed. Seems, that there are only about 6 people with write 
access right now. (Source: 
http://sourceforge.net/project/memberlist.php?group_id=24919).

Anyway, make sure to keep the daily CVS tarballs generated by sourceforge as 
outlined in

http://sourceforge.net/docman/display_doc.php?docid=768&group_id=1#top

Look for 'Nightly CVS Tarballs' on that page.

- -- 


Regards

Thomas Baumgart

home: thb at net-bembel.de              WWW: http://www.net-bembel.de/
Check the KMyMoney2 open source project at http://kmymoney2.sf.net/

public key available at http//www.keyserver.net under thb at net-bembel.de
GPG-FP:  E55E D592 F45F 116B 8429   4F99 9C59 DB40 B75D D3BA
- -------------------------------------------------------------
A crash turns an expensive computer into a simple stone!
- -------------------------------------------------------------

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: Visit S.E.S.A. AG at http://www.sesa.de/

iD8DBQE9N+NWBEBpXFzzndgRAl1UAJ9NflJjmgbyk5RVtTx5HqT4CstWdgCeLfOU
uoPnFKdTOO0no13sYb9KRbI=
=tkIU
-----END PGP SIGNATURE-----