[Yum-devel] metadata parser in C

Konstantin Ryabitsev icon at fedoraproject.org
Wed May 10 14:18:27 UTC 2006


seth vidal wrote:
> Also thanks for working on this, it looks cool. I'm going to give you
> some initial thoughts and since I've just jumped out of the shower some
> extended thoughts, too. So take all of it inclusively and we'll see what
> comes out, okay?

Since metadata parsing is in no way related to the rest of yum's 
operation (yum works with post-parsed databases), I think this is a very 
sane thing to do. The slowdown in case of python-cElementTree was the 
fact that we had to do the c-string -> python-string -> c-string 
hoop-jumping (even worse in case of libxml, since then it also insisted 
on converting unicode objects into strings). If we never leave C to 
parse and store metadata, the speed-ups should be tremendous, which is 
what has been demonstrated.

I don't think we should worry about yum having bits in C, especially 
since it only concerns metadata parsing. It's a visible enough project 
these days that finding a maintainer for this (probably very small) 
piece of code shouldn't be in the least problematic.

Regards,
-- 
Konstantin Ryabitsev
McGill University WSG

Mal: "You're servicing crew now?"
Inara: "In your lonely and pathetic dreams."
				--Episode #2, "The Train Job"



More information about the Yum-devel mailing list