Fixes issue reported on the mailing list here: https://lists.archlinux.org/pipermail/aur-general/2022-March/036833.html

Thanks to Henry-Joseph Audéoud for diagnosing the issue
https://lists.archlinux.org/pipermail/aur-general/2022-March/036836.html

Also update the event variable to use the local copy instead of the
deprecated global version
https://stackoverflow.com/questions/58341832/event-is-deprecated-what-should-be-used-instead

Signed-off-by: Kevin Morris <kevr@0cost.org>

Straight up bug.

Closes #324

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

...This was an obvious bug in hindsight. Apologies :(

Closes #314

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

This was causing some false negative errors in the update process,
and it clearly not correct -- oops :(

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Additionally, respond with a 503 if the var is not set when
/metrics is requested.

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

In addition, treat package search by keywords as lowercase.

Closes #296, #297, #298, #301

Signed-off-by: Kevin Morris <kevr@0cost.org>

Somehow, many aur.al records of PackageVotes do not have a valid VoteTS
value. This migration fixes that issue by setting all null VoteTS
columns to the epoch timestamp.

Signed-off-by: Kevin Morris <kevr@0cost.org>

Closes #284

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

We need to query for this after we've applied all filters.

Signed-off-by: Kevin Morris <kevr@0cost.org>

The previous behavior was carried over from PHP. It has been requested
that we use the true defaults when rendering the default form, making
search a bit more sensible.

Closes #269

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

This was not including other parameters that should be persisted for
users.

Signed-off-by: Kevin Morris <kevr@0cost.org>

These were being produced with the db state before the flag was set,
which is not what should be done for flag notifications, as the
notification contains data about the comment and the current flagger.

Closes #292

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

We were redirecting in some error-cases, which this commit sorts out:
- package count == 1 and package base name != package name
- was redirecting to {name} and not the only associated Package

Now, when we have a package base name that mismatches its only
package, we display the package base page. Otherwise, we redirect
to the first package's page.

Closes #282

Signed-off-by: Kevin Morris <kevr@0cost.org>

Closes #283

Signed-off-by: Kevin Morris <kevr@0cost.org>

Closes #279

Signed-off-by: Kevin Morris <kevr@0cost.org>

Signed-off-by: Kevin Morris <kevr@0cost.org>

Closes #278

Signed-off-by: Kevin Morris <kevr@0cost.org>