[Rpm-metadata] Another suggestion

[Daniel Veillard]

On Sun, May 30, 2004 at 05:37:04PM -0400, seth vidal wrote:
> On Sun, 2004-05-30 at 17:54 -0300, Gustavo Niemeyer wrote:
> > > Why is that particularly valuable? The xml parsers I've seen all read
> > > through the gzipping transparently and I've not noticed a speed hit by
> > > reading it as a gzipped file.
> > 
> > How many packages you've done your measurements with? What about with
> > bzip2?
> 
> I've only tested with libxml2 and pyXML. libxml2 being significantly
> faster than pyXML.
> 
> > > Where's the advantage of storing it uncompressed?
> > 
> > Speed. Uncompressing it only once prevents paying the same price
> > everytime you read them.
> 
> fair enough. I'd love to see if that's a signficant hit or not but after
> discussing on irc I'm willing to accept it at face value.
> 
> unless anyone has an objection I'll store the uncompressed information
> in the repomd.xml file as well.
> 
> by information I solely mean:
> sha/md5 checksum
> timestamp

  remember that one of the design goal was to minimize the amount of 
metadata, if there is redundancy or special needs then they should
be done as extensions in another namespace.

Daniel

-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard at redhat.com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/