> .
>
> Then there's no sense just passing the whole thing through as a
> transaction to rpm, right?
no, but instead of failing at that point yum could "think for you" and
prune out the already installed or fully updated ones to let the
operation continue - and politely warn you to stderr that this is
already installed.
> On my box it's more like 30 seconds, which is why I brought it up to
> begin with.
really? Do you have A LOT of headers to process?
Once I get 2.0 out final I'm planning on slipping timeit.py into things
and seeing where all the time is being spent.
> I wouldn't bother with this issue then. Work on something that's
> actually important, or that you'd enjoy.
hmm. something I'd enjoy. There are 2 or 3 "cool things" on the todo
list that I'd love to play with but they have to be on the back burner
for the moment.
cool things in order of priority to me:
- depresolution w/o using ts.addInstall(hdr, 'a')
- make yum be able to install into a buildroot
- make yum be able to read a file for the list of pkgs
ie: something like the %packages section of a ks.cfg
- write a data-dump mode for yum to spit out all the various info about
the system
btw - if anyone out there wants to take a stab at any of these things
I'd be interested to see code :)
-sv
