[Ifeffit] Re: Horae-058

[Matt Newville]

Hi Bruce,

> > I tried downloading the latest tarball using your horae-update
> > script.  It download a file with the right name, but that only
> > contains what appear to be sourceforge related web links.  In other
> > words horae_update is broken (much like what happened before).
>
> SourceForge redesigned their web site.  The screen scraper in
> horae_update no longer works.
>
> I am hesitant to just fix the current code -- it was a hand-built
> solution and, obviously, much too fragile.  A better solution would
> require either using better html parsing tools from CPAN or figuring
> out how to gather information from SourceForge without needing to
> scrape their web pages.
>
> I am open to any suggestions or offers of coding help.

Would it make sense to use sourceforge's cvs repository to download
horae?   I think that would be much simpler and more stable to script.
   And if a user kept the downloaded tree on their machine,  it would
also be much less stuff to download for an update.    The downsides
are that the sourceforge CVS does not have the most reliable uptime,
and it probably really means going to a single machine in the US.

The other alternative is to just use cars9.uchicago.edu.   Is there
really a noticeable improvement in accessing the files through
sourceforge than directly from cars9.uchicago.edu?    When I go to use
sourceforge and choose to download a file from Europe or Asia the
download itself seems to go pretty quickly compared to navigating
sourceforge itself.    Sadly, I don't know of any better alternatives
to sourceforge.

--Matt