Re: [Chicken-hackers] Egg filename versioning

[Leonardo Valeri Manera]

On 08 Feb 2008 13:16:48 -0200, Mario Domenech Goulart
<address@hidden> wrote:
> Hi Graham and folks,
>
> On Thu, 7 Feb 2008 14:19:16 -0500 "Graham Fawcett" <address@hidden> wrote:
>
> > On Feb 7, 2008 1:18 PM, Leonardo Valeri Manera <address@hidden>
> > wrote:
> >
> > > On 07 Feb 2008 15:49:56 -0200, Mario Domenech Goulart
> > > <address@hidden> wrote:
> > > > Hi Leonardo,
> > > >
> > > > On Thu, 7 Feb 2008 16:42:33 +0100 "Leonardo Valeri Manera" <
> > > address@hidden> wrote:
> > > >
> > > > > To make it all actually work though, we need to be able to
> > > > > differentiate the different egg versions by filename ... that's not to
> > > > > say it has to be the actual *VERSION* - if that's too complex to do,
> > > > > the release date will do just fine:
> >
> >
> > How about something like this?
> > http://cleo.uwindsor.ca/cgi-bin/gentoo-eggs/
> >
> > It is just a proof of concept, please don't point the gentoo universe at
> > this link. :-)
> >
> > It fetches the "tree" of metadata from the Chicken site, generates a list of
> > links, and proxies the download of eggs so that the filenames contain the
> > necessary timestamp.
>
> Great work!  To avoid inconsistencies, would it be possible/feasible
> to fetch the metadata [*] at each request?
>
> [*] http://www.call-with-current-continuation.org/eggs/3/repository
> (used by chicken-setup).
>
> PS: you get an internal server error if you strip the last slash from
> the URL (http://cleo.uwindsor.ca/cgi-bin/gentoo-eggs).
>
> Thanks, Graham! Neat solution!

I endorse the above comment!

Also, good point about the metadata.

As I said, once these ebuilds hit the main tree the eggs will be
fetched from our mirrors apart from the occasional case of a
badly-configured user installation (doesn't happen often with
schemers, I think), and we'll be implementing dependencies internally
anyway, so we should not be putting too much load on the servers -
hopefully :D

Cheers!
Leo