|
Boost : |
From: Aristid Breitkreuz (aribrei_at_[hidden])
Date: 2005-10-15 06:09:21
Andreas Pokorny <andreas.pokorny <at> gmx.de> writes:
>
> Hi,
> Some weeks ago someone metioned that we need a C++ variant of
> CPAN or CTAN. The idea is really great.
Yes!
> But I have no clue where
> we would start. How should that service look like, and work internally?
We need a comfortable web frontend *g*.
> I think we need some identifcation service for authors, maybe some
> access control on the packages stored in the serivce. The client frontend
> will need something to select mirrors, track dependencies to other packages.
Let me note that using BitTorrent for that would reduce the cost of such a
system.
> An auto build, intstall and test feature would be nice too.
But pretty hard.
> As Martin Wille just mentioned, something with some of these features
> already exists. Gentoos portage [2], written in python handles,
> dependencies, integrates a mirror selector, automatically build installs
> and uinstalls packages fetched as source code. The package scripts are
> shared using rsync[3]. The package source files are automatically
> distributed into their mirror system. The only missing thing is the
> author identification and access control part. Gentoo "works" with
> an indirect system, requiring authors to provide scripts, or users to beg
> for packages in their bugzilla system.
Well, surely ideas can be stolen from Portage.
> Another interesting system could be provided by monotone[4], a very nice
> distributed version control system. We could use monotone to replace
> rsync, in the suggestion above. Monotone provides a secure system to
> get the build scripts anonymously, and allows identifying authors. The
> distributive character of monotone allows very intersting use cases, in
> which authors are able to update and provide new versions of their
> software.
I do not think monotone is what we need here. The problem is that you cannot
checkout from a monotone server without downloading all prior versions. Monotone
is really made for distributed development and not for such services.
> Regards
> Andreas Pokorny
Sincerely, too,
Aristid Breitkreuz
>
> _______________________________________________
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost
>
>
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk