[Savannah-dev] Re: Savannah issues

[Mathieu Roy]

Paul Fisher <address@hidden> said:

> Since anoncvs has been back up and running, it seems that there are at
> least a few clients out there that are not playing nicely.
> 
> There is one CVS process that has been running for about 12 hours (job
> 13857).  Seems rather unlikely that a checkout would take 12 hours,
> but maybe it's legitimate, and in another 12 hours the process will be
> gone.
> 
> There's a new stable release of CVS out (1.11.6), and I'm pondering
> upgrading CVS on savannah.  One of the bugs that appears to be fixed
> in 1.11.6 (that isn't in our current version of CVS):
> 
>     * There was a bug where certain error conditions could cause the
>     server to go into an infinite loop.  There was also a bug that
>     caused a compressed connection from an older client to hang on
>     shutdown.  These bugs have been fixed.
> 
> And of course this brings up the NGROUPS_MAX issue again and how we
> need to have custom versions of important software on savannah.
> 
> I'm not confident that the NGROUPS_MAX issue will be resolved in time
> for linux 2.6.  Has any progress been made on using extended
> attributes instead of groups for access control? 

No.

> If no one has looked into extended attributes yet, I'm going to try
> to squeeze in some time over the next few weeks to at least see what
> work would be required to make it happen.

It would be good to list what need to be done.

> 
> 
> Looking further toward the future, are there any plans to move to
> GForge 3?

I checked what it provides.
My personal opinion is we have in Savannah features missing and not
planned in gforge and gforge is designed in the SourceForge way, what
we are currently undoing step by step with the Savannah code.
(for instance, we made Savannah w3c compliant, gforge people apparently
don't care, because it's far from it)

Also, one more personal opinion, I have the feeling that enough people
from gforge do not understand freedom in computing like I do and I
would not like very much being bound to them.

So I'm personally against gforge as a solution for savannah.gnu.org,
in the current state of things. But if the FSF want to switch over
gforge, it can happens (without me, but that's not a problem, I can be
replaced easily).

I think that 2 differents project for a same general purpose are a
good option when enough differencies exists between them, and my
personal feeling is Savannah is really different from gforge.

I spent hours working on the Savannah codebase and think we can
greatly enhance it instead of trashing it and switching to something
that have the same issues that we fixed at first on savannah. 
I would be a lot more favorable to the implementation of free
interesting stuff existing in gforge into Savannah than abandonning
our fork.
I'm not saying that gforge is bad, but Savannah has another history
and in the state of thing, I do not think it would be wise to erase
it.
So, to me, gforge is not more the future than savannah can be. Gforge
is not the continuation of Savannah and Savannah is no longer just an 
installation of SourceForge on a GNU computer.


Regards,

-- 
Mathieu Roy
 
  Homepage:
    http://yeupou.coleumes.org
  Not a native english speaker: 
    http://stock.coleumes.org/doc.php?i=/misc-files/flawed-english