[TYPO3-dev] The neverending sf story

[Dimitri Tarassenko]

Michael,

On 3/11/06, Michael Scharkow <michael at underused.org> wrote:
> Sven Wilhelm wrote:
>
> > ok, i'm not on the core list, so I can't recognize such discussion.
> > But if such points are discussed on the core (a complete closed list),
> > it's perfect for the others outside to see nothing early of the there
> > discussed topics :)
>
> AFAIK the core list is publicly readable, so only participation is
> somewhat limited ;)

I think a discussion here is also beneficial since SF not only hosts
the core but also the typo3xdev, and if the core was to move someplace
along with the Mantis bugs for some extensions, it would be logical to
move the typo3xdev too.

Also, Trac's authentication is very flexible, so we can finally bring
the 4 different logins used by TYPO3 developers together
(bugs.typo3.com, typo3.org/TER, SF and, potentially, Wiki). If this
came true it would be just awesome!

Also, as a possibility: on the project that I am managing now we are
replicating an off-site SVN repository into the one attached to Trac,
and it works flawlessly. Michael Stucki is right in his reservations
against the idea of leaving SF alltogether, so, what could be done is:

a) build a Trac install with authentication through typo3.org/TER user DB
b) migrate Mantis bug history there
c) migrate Wiki
d) build a replication with the SVN repository in SF, either
   d1) a slave copy for referencing from bugs with no SVN checkins/checkouts
   d2) a master copy that would "push" checkins to SF, with anonymous
checkouts through SF only.
e) continue serving the files through SF file release system. Someone
raised concerns about "clicking several times" - this can all be
solved by "resolving" these several clicks on the server side at this
new site. One of the projects I am working on and off is a YUM/apt
repo interface over the SourceForge file release system and this is
what I do there.

> > 2. For setting up a testing environment, preparing a rock solid tested
> >    environment I would setup a virtual server.
> >    For final serving the server should have enough RAM, but this one
> >    could be one the list when testing of point 1. is finished.
> >    As I told I would sponsor money therefore.
>
> I'm just setting up a dedicated XEN instance on our $FATSERVER now for
> testing, as I'd like to test svn 1.3 features and our webserver runs
> only Sarge...

I'd chip in on this too - I can see a lot of good out of this move, so
I guess when the time comes and you're ready you can just start a
fundraiser here and in the core list - good enough dedicated servers
are ~$100-$120/mo. now.

--
Dimitri Tarassenko