[arch-haskell] http-conduit

[Ramana Kumar]

On Mon, Sep 24, 2012 at 11:37 PM, Fabio Riga <rifabio at gmail.com> wrote:

> In data 20.09.2012 17:34:15, Nicolas Pouillard ha scritto:
>
>> Thanks it seems to work great!
>>
> Thank you! Please note that this is a test. So expect this to break in
> some way.
>
>>
>> Can you tell how easy to maintain a repository (cblrepo) on top of the
>> archhaskell one. I did thought about forking habs but it is a waste of
>> time. However building another repo on top of it might provide with a
>> nice distributed nature allowing us to maintain the package we care the
>> most about.
>>
>> Did you took special steps, how do you merge the cblrepo from the main
>> habs?
>> You turn them from RepoPackages to DistroPackages right?
>>
>
Why was this - what looks like a very interesting - message not posted to
the list?


>  This was exactly my idea. There's no need to fork the excellent work
> made by Magnus and others. I made a new repo using cblrepo. I inserted the
> same GhcPkg of habs. Then I made a script for adding new packages, tracking
> all dependencies using cabal install, and check them with the habs repo. If
> packages are there, the script uses cblrepo to add them as DistroPkg, else
> they are inserted as repo.
>
> The script is far from being completed and usable, and I haven't had
> enough time to clean, document and publish it. I would like to make it easy
> to make a new repository tracking many others, not just one. So to make, as
> you told, a distributed system.
>
> The main problem I encountered (and everyone using [haskell-extra] will
> do) is that when a package in [habs] is updated and I have a package
> installed that depends on it, pacman will refuse to update the system until
> [haskell-extra] is updated as well. This is unavoidable, as I need the new
> package in [habs] before updating [haskell-extra]. The alternative would be
> to recompile all updated packages in [habs] AND all affected package in
> [haskell-extra], and I won't do this! Another way could be to have updated
> packages in a private repository for a couple of days, so my repository
> (and others as well) could have the time to keep in sync. With the latter
> solution, we could easily merge the resulting packages in one repository.
>

If there is a single repository that merges all the real distributed
repositiories, that is the only one that needs to be "public".
I suggest renaming [haskell] (to haskell-core or haskell-init), and
creating the merging repo with the name [haskell] right away, and see if it
works for merging current [haskell] and current [haskell-extra].
I hereby offer my server to get this up and running initially (we can move
it elsewhere as necessary later).


>
> I hope this clarify a little what I'm doing. Any suggestion is highly
> appreciated.
>

My suggestion above to make the merged repo now, and the meta-suggestion to
refocus on being inviting so we can easily get more people/resources/time
from the community.


> Fabio
>
> ______________________________**_________________
> arch-haskell mailing list
> arch-haskell at haskell.org
> http://www.haskell.org/**mailman/listinfo/arch-haskell<http://www.haskell.org/mailman/listinfo/arch-haskell>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.haskell.org/pipermail/arch-haskell/attachments/20120925/51ad7c36/attachment-0001.htm>