Age | Commit message (Collapse) | Author |
|
|
|
|
|
|
|
|
|
|
|
Links to patches get outdated ever time we move it to new commitfest.
|
|
|
|
Now it forwards everything after the slash so that links like the
following work:
/open/new/
/open?author=-3
|
|
django-selectable is no longer maintained, and will cause issues in
newer versions of django.
selectize is what we use in the pgeu-system codebase today, so copy the
handling over from there.
|
|
The previous commit removed the refusal, but missed adding the allwing..
|
|
Discussion: https://postgr.es/m/20211001173131.GA13090@ahch-to
|
|
|
|
|
|
These clearly had little to do with reality since they would return
duplicate entries if a patch was in more than one cf..
|
|
This is particularly interesting towards the end of a cycle where it can
be used to flag patches that are not intended for the current version
but still needs review.
|
|
|
|
|
|
|
|
By popular (?) request
|
|
Instead use the value from the settings file, which is what's actually
used to send.
|
|
|
|
Same fix as the previous one for status, but applies to the other two
filter-by-dropdown fields.
|
|
The comment said we should ignore the status if int() on it fails, but
to do that we have to fail it before we actually add the WHERE clause.
Thus, reorder the code to do so.
|
|
Accidentally broken when query was rewritten
|
|
We've long gone past what the django ORM can handle and got into
hundreds of queries for large commitfests. And it's not really that hard
to rewrite as SQL, even though we have to dynamically build the WHERE
clauses. So do that.
|
|
Per discussions at the developer meeting, show the number of CFs a patch
has been on, and make it possible to sort on this value.
|
|
Previously we'd throw a 500 internal server error, and cause error
logging to happen. This became evident when scripts trying to SQL-inject
the site started hitting us..
Instead, just ignore any filters that pass non-integer values into
integer fields completely.
|
|
This allows the archivs server to ping the CF app to pick up updates to
mailthreads quicker.
|
|
This is rquired as part of the move to a newer Django, and works fine on
older versions too
|
|
Previously we would set the Cc header of the email, but not actually
send a copy to the user. That way if the user was not subscribed to the
hackers email, they'd end up not receiving the email at all, which
somewhat defeats one of the original points of adding the Cc in the
first place.
Spotted by Peter Geoghegan
|
|
Patch in 2e41b31654b80aeb3e6037fc0b31422c951040c7 only handled From and
missed the ohher tields. To make this cleaner, move the escpaping code
into the UserWrapper class.
Reported by Dagfinn Ilmari Mannsåker, but not using his patch
|
|
We need to treat them as structured and escape only the name, not the
email part.
Reported by Dagfinn Ilmari Mannsåker
|
|
This adds the Cc field when using the Review or the Comment function on
the website, and will add a list of all listed authors of the patch to
this. Should help people highlight it in their inboxes.
Suggested by José Luis Tallón
|
|
So a patch that's in "ready for committer" status remains there.
When doing this, also refuse to move any patches that are not either
"waiting for review" or "ready for committer". They will have to have
their existing status changed first, and then be moved.
|
|
This makes it possible to receive notifications (via email) when a patch
has been updated.
It's possible to subscribe to notifications on a specific patch. It is also
possible to set the user profile up so that one receives notifications for
all patches where one is author, reviewer or committer on.
Notifications are, of course, only sent when other people make modifications
to the entry.
|
|
|
|
|
|
This is an incorrect state, but we shouldn't crash anyway. Instead show
an error message and redirect to the root page.
Reported by Bruce Momjian
|
|
Exceptions will not show the actual error message on the site, only
send an email to the admins with it. Thus, anything that can normally
happen due to a user error should not use that. Instead of that, use
the messages framework so it renders on the next page.
|
|
|
|
|
|
feedback"
Returned with Feedback will now actually close the patch and *not* move it to
the next CF. Since it used to do that, all existing Returned with Feedback patches
are changed to Moved to next CF which does the same thing that Returned with
Feedback used to do. And of course, Moved to next CF is a new option available for
each patch from now on.
|
|
Without this, once a patch was closed (rejected/committed etc),
it was no longer possible to undo that and reopen it.
We now allow this to be done in the very last commitfest a patch
is on. We cannot change it on a previous commitfest, as that could
end up with the patch being open in two CFs at the same time, but
there is no reason it shouldn't be doable in the last one.
|
|
Quick overview of how many patches are in each status, as exist on
the old app.
|
|
Suggested by Stefan K
|
|
Instead include the actual username. Makes it easier to work with in
the history views, and is also probably more proper english.
|
|
This needs to redirect to the login page and back, so the user is
actually logged in when attaching. This was only partially ipmlemented
before, it turns out.
|
|
Without using .as_string() on a message, the default is to generate
a string including the envelope header used in a unix mbox. This would
of course generate completely broken messages when passed to smtp.
|
|
Each user can add a secondary email (well, more than one) and then pick
one of those when sending email.
Addresses are validated by sending a token to the newly added address,
with a link to click to confirm it. Only a fully confirmed address can
actually be used.
|
|
This somehow got dropped from the old site in a feature branch that
was never merged...
|