"CPAN" SIG?

[Brad Clements]

I'm seriously thinking about using a modified CODA system on Linux to create
a truly distributed object database for Zope, which could be a good basis
for a distributed "Python-Pan" system.

The other issues such as installation tools still need to be considered, but
I'd sure like to tackle this problem using CODA or perhaps AFS when its
open-sourced this month by IBM.



--
Brad Clements,
bkc at murkworks.com
"D'Arcy J.M. Cain" <darcy at vex.net> wrote in message
news:8qva89$1rft$1 at news.tht.net...
> Tom Loredo <loredo at spacenet.tn.cornell.edu> wrote:
> > Vaults is great.  However I strongly feel there is a need, if not
> > for CPAN, then at least for mirrors of Vaults.  As someone who has
> > just written a grant proposal seeking support for a Python
> > project, I find it worrisome that Python's main source of info
> > about 3rd party libraries is a single site.  In particular, Vaults
> > has been incredibly unstable in the last month or so; a web page
> > supporting my proposal links to it and I am just praying that
> > when the reviewers visit it, it's not one of the days that clicking
> > on the Vaults link returns an error message.  There have been
> > many days like that lately.
>
> There should be fewer lately.  Let me tell you what has been going on.
>
> Recently we did a major upgrade on Vex.Net.  All third party packages
> were upgraded and rebuilt as well.  Since then the VoP (which is built
> on Python and PostgreSQL) has been giving us troubles.  Specifically,
> the database engine kept crashing.  We finally figured out that it was
> the VoP database that was the problem and we created a second instance
> of the engine to handle it.  This has confirmed that the problem is
> related to the VoP.  We are now watching it carefully and are in the
> process of writing a script to monitor and restart as necessary.  I
> have also had Tim send me the schema for the database so that I can
> look it over and possibly help in case the problem is there.  I will
> also be helping him go through the code to see if we can find where
> it crashes.  Meanwhile, the Vaults should be up most of the time now.
>
> > The collective efforts of Python programmers are too vital a
> > component of the case for Python to be subject to this kind
> > of single-point failure.  We need something broader and more
> > stable.
>
> This, of course, is a different issue.  Have you contacted Tim about
> doing some mirroring?  As long as you can install PostgreSQL and PyGreSQL
> you should be able to run the same code.  You may even be able to help
> debug the current problems.
>
> --
> D'Arcy J.M. Cain <darcy at vex.net>           |  Democracy is three wolves
> http://www.vex.net/                        |  and a sheep voting on
> +1 416 425 1212     (DoD#0082)    (eNTP)   |  what's for dinner.