On Mon, 2005-03-07 at 12:04 +0800, Tsai Li Ming wrote: > seth vidal wrote: > >>I was going to look at it tomorrow to see if I can reimplement mdparser > >>in cElementTree. It's hard to say which one is faster without having > >>side-by-side running code, but cElementTree site claims that it smokes > >>libxml2, plus it's a clear win in terms of the APIs: as opposed to > >>libxml2, the author of cElementTree has some idea about what makes good > >>Python bindings, so you don't feel like flossing with barbed wire when > >>writing code in it. > > > > > > I did some import tests using cElementTree last night, it's, umm, big, > > memory-wise. reading in filelists.xml.gz for only rawhide ate up 120M > > > > I think with the FancyTreeBuilder it can be made a lot smaller, not sure > > yet. > > > > -sv > > Hi Seth, > > You might want to check with effbot. cElementTree is suppose to be > faster and takes up less memory than the rest. > we discussed it some more on yum-devel - it's just the repoview was not using iterparse() to handle the xml file - so it was reading the whole 100M xml files into memory :) and you're right cElementTree is much faster than libxml2 from what we've seen so far. -sv
