Dist. Object Directory (was: Re: An FTP based Python Module repository)

chris patti cpatti at atg.com
Fri Feb 4 16:42:06 EST 2000

Aaron J Reichow <reic0024 at ub.d.umn.edu> writes:

> On 4 Feb 2000, chris patti wrote:
> > Another big win in this is that it eases the automated retrieval and
> > installation of modules, in the Perl world we have CPAN.pm which lets
> > you say things like:
> > 
> > install frobnitz
> > 
> > And it will search the archive, find the latest version, download it
> > and install it to your local Perl installation.
> And to me, a similarily interesting idea would be to have a server with a
> number of mirrors, that could host remote objects using dopy (alpha code,
> but easy to use IMO) or fnord.  This probably isn't the most practical of
> all things, but the idea fascinates me. 
> Aaron

Well, it sounds cool but it's the wrong level of abstraction for the
problem :)

All we need is a single archive where all the Python modules in 
existence (that their authors want published) live.

The server generates a list of what modules live where, and so when 
a client asks for "xmlrssparse" the server knows that this is really
stored in authors/reiley/Xml-Parser-With-Rawpberry-Frosting.tar.gz
which unpacks and installs the xmlrssparse module and any modules it
depends upon.

How would an object broker approach help this situation?

Just wondering.


Chris Patti| \ Art Technology Group \ 617-386-1649 \ cpatti at atg.com

More information about the Python-list mailing list