seth vidal wrote:
I gave a talk on yum at fudcon. It was fairly well attended. I decided
to preempt the 'yum sucks' portion of the talk by putting up a white
board with 'yum problems' written across it then asking for what
problems people have with yum.
Here's the list I got, some of them were suggested by people with less
than complete sincerity:
- it's not apt :)
Well, this could be fixed quite quickly ;)=
- make it faster/smaller/______er
I think we already did a lot of progress in this area during the last 6
months. There is probably another factor 2 to gain but further optimizations
will get harder and harder. There are also some situation where we currently
start again from scratch where this could be avoided. This might be easy and
worth to fix.
But recent experiments with large transactions show that yum is quite
efficient while rpm(lib) uses up a lot of time and memory (2 hours, 1.2 GB
vs 2m30s 200MB). This means further optimizations in the yum depsolver will
feel good but won't really help much (at least for those cases).
As a result of this I am planning to have a look into rpm and to try to fix
these issues during this year. But don't expect any miracles soon.
- the search faster and search having an issue on generic searches I
think I know how to beat up a bit more. Sorting the search results is
great but I think we need to either add a LIMIT operation to the search
results of some kind.
Has someone already experimented with sqlite full text indices?
(http://www.sqlite.org/cvstrac/wiki?p=FullTextIndex) I am not sure it it is
worth creating all the infrastructure just for using it in yum but it might
be worth evaluating this.
Florian
_______________________________________________
Yum-devel mailing list
[email protected]
https://lists.dulug.duke.edu/mailman/listinfo/yum-devel
