| Age | Commit message (Collapse) | Author |
|---|
|
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Put an actual NULL in the database and handle it for both display and
import. Also add a migration to clean up any bad data we currently have in
there. Fixes FS#17144.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Signed-off-by: Evangelos Foutras <foutrelis@gmail.com>
[Dan: made a few other small touchups]
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
We didn't sanity check the length of the DB set, so if it was zero we would
blow up. Add a sanity check and also limit the whole thing to only apply if
there are > 20 packages in a given {repo, arch} combo.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
With community testing, we want to ignore the ratio check failure there as
well. Clean up the whole check a bit and store the percentage in a variable
as it is useful in both checks and for output purposes.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Fixes FS#14270.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
|
|
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
|
|
We had a bit of a disparity between the add and update sections in reporead,
causing issues like FS#12400 to only manifest themselves on updated packages
rather than brand new ones. Unify the code into a function that does the
updating so this doesn't happen again.
Signed-off-by: Dan McGee <dan@archlinux.org>
Signed-off-by: Dusty Phillips <buchuki@gmail.com>
|
|
|
|
Commit 789b5445cf originally set out to fix this problem but was not
complete. When doing an update of [extra], the code still did > 2000
database queries because of an ill-placed call to QuerySet.get(). By using a
dictionary lookup, we reduce the number of database queries to a measly 4
when updating extra and no new packages are present.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
|
|
|
|
Rather than load one package at a time and put unnecessary load on the
database, load them all at once upfront since we are going to need 99% of
them anyway. This shifts the burden of work from the database to the python
script itself.
Signed-off-by: Dan McGee <dan@archlinux.org>
|
|
|
|
|
|
|
|
|
|
This patch adds URL to the fields to be updated on package update.
Path is a fix for the following scenario:
- a package is added at one point.
- later a url is changed or added.
- when the package is updated, the following fields are updated:
pkgver, pkgrel, pkgdesc, needupdate, last_update
There was an original assumption, that some fields were very seldom changed,
and it would not have make sense from a performance standpoint to update the
fields that never changed, since the update mechanism was written using the
django orm..which is 'chatty'.
NOTE: It is rather pointless to do a compare to test if the field has changed.
It is more performant to simply update the field regardless.
|
|
|
|
|
|
- Fix for reporead when description is empty.
- Fix for settings.py. Middleware wasn't found (changed import name)
- Clarity change for dashboard.
|
|
|
|
|
|
Chop so we don't throw mysql errors when we try to commit.
|
|
Added them to the django admin interface as well.
|
|
|
|
|
|
|
|
reporead uses the same model files, and needs access to the db anyway.
|
|
|
|
Special Note
Prior to git import, approx 90% of the code was done by Judd Vinet. Thanks Judd!
|
