Commit 30eaed0378 (#15001) incorrectly
inserted a different section between the anchor and the heading.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
The auth attempt rate limit is quite low (on purpose), so this can be a
common scenario where a user asks their admin to reset the limit instead
of waiting. We should provide a tool for administrators to handle such
requests without fiddling around with code in manage.py shell.
Calling `email.save()` is only needed if we altered `email.address`;
it is unnecessary if we called `email.users.add(...)` which will have
done its own INSERT.
This fixes two bugs: the most obvious is that there is a race where a
ScheduledEmail object could be observed in the window between creation
and when users are added; this is a momentary instance when the object
has no users, but one that will resolve itself.
The more subtle is that .save() will, if no records were found to be
updated, _re-create_ the object as it exists in memory, using an
INSERT[1]. Thus, there is a race with `deliver_scheduled_emails`
between when the users are added, and when `email.save()` runs:
1. Web request creates ScheduledEmail object
2. Web request creates ScheduledEmailUsers object
3. deliver_scheduled_emails locks the former, preventing updates.
4. deliver_scheduled_emails deletes both objects, commits, releasing lock
5. Web request calls `email.save()`; UPDATE finds no rows, so it
re-creates the ScheduledEmail object.
6. Future deliver_scheduled_emails runs find a ScheduledEmail with no
attending ScheduledEmailUsers objects
Wrapping the logical creation of both of these in a single transaction
avoids both of these races.
[1] https://docs.djangoproject.com/en/3.2/ref/models/instances/#how-django-knows-to-update-vs-insert
Only clear_scheduled_emails previously took a lock on the users before
removing them; make deliver_scheduled_emails do so as well, by using
prefetch_related to ensure that the table appears in the SELECT. This
is not necessary for correctness, since all accesses of
ScheduledEmailUser first access the ScheduledEmail and lock it; it is
merely for consistency.
Since SELECT ... FOR UPDATE takes an UPDATE lock on all tables
mentioned in the SELECT, merely doing the prefetch is sufficient to
lock both tables; no `on=(...)` is needed to `select_for_update`.
This also does not address the pre-existing potential deadlock from
these two use cases, where both try to lock the same ScheduledEmail
rows in opposite orders.
No codepath except tests passes in more than one user_profile -- and
doing so is what makes the deduplication necessary.
Simplify the API by making it only take one user_profile id.
On mobile, when the sidebar is toggled, the following three issues
are encountered:
- When none of the sidebar menus are expanded, the sidebar has no
scrollbar, which is expected. But if you scroll, the background
content scrolls, which is a bug.
- When some of the sidebar menus are expanded such that the content
overflows and is "scrollable", once you get to the end of the
sidebar content, the background content keeps scrolling in a weird
way.
- If the mobile screen is wide enough, if you scroll the sidebar
content, it scrolls as expected. But if you move the pointer to
the side of the background content that is still visible, you
can scroll the background content even though it should be fixed.
This commit fixes all of the above issues.
This fixes a bug where email notifications were sent for wildcard
mentions even if the `enable_offline_email_notifications` setting was
turned off.
This was because the `notification_data` class incorrectly considered
`wildcard_mentions_notify` as an indeoendent setting, instead of a wrapper
around `enable_offline_email_notifications` and `enable_offline_push_notifications`.
Also add a test for this case.
None results in an uninitialized image (that happens to be transparent
most of the time); we want to explicitly initialize the image to
transparent.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
While the STREAM_LINK_REGEX and STREAM_TOPIC_LINK_REGEX
identifies the stream and topic mentions in the content
correctly (tested by printing out the matches), the
stream/topic mentions are still not linked to the
corresponding streams/topics for imported messages, as
a `zulip_message` instance is required for linking these
mentions to actual streams/topics (see `StreamPattern`
class in `markdown/__init__.py`) which is not provided
while processing the markdown for imported messages.
This commit updates both the stream-level and realm-level message
retention setting to use 'unlimited' instead of 'forever' to set
message retention setting to "retain messages forever".
We incorrectly include many realm settings in the data section of
'realm/update_dict' schema. It should only contain the settings
related to message edit, realm icon, realm logo and authentication
methods and not other settings, becausea all the other settings send
'realm/update' event and not 'realm/update_dict' event.
This commit only removes 'message_retention_days' and others will
be removed separately.
Earlier copy to the link present above the composebox (which
appears when a message is sent outside the current narrow), it
used to provide the link to the home page which is fixed now.
Now, copy to the link gives the link of the narrow where message
is sent.
When one resizes the window and tries to switch to the vertical
sidebar menu, the CSS transition flickers in and out. This is
less than ideal.
This commit implements a solution to this problem. The solution is
inspired by the following helpful article:
https://ishadeed.com/article/layout-flickering/
We found a bug in our top-level nav on certain Samsung phones where
tapping on the dropdowns didn't do anything. As it turns out, trying
to figure out whether touch is enabled on a device using CSS media
queries such as `(hover: none)` is very tricky business! While it
may work on some devices, it may break spectacularly on others.
This commit presents a potential solution to this. Media queries
about widths are a lot more device-independent than hover media
queries. So, this is what we do now:
- We fire click events regardless of whether the top-level nav is in
the vertical sidebar orientation or the horizontal orientation.
Since we are no longer dependent on hover media queries for click
events, this fixes the mobile bug mentioned above because taps or
clicks will always work regardless of device.
- The mouseover/mouseout events are only fired when the header is
in the horizontal orientation and the primary input mechanism
supports hovering over elements. This allows us to support hovering
over menus on desktop. However, since mouseout/mouseover events are
irrelevant for mobile, we are fine on that front since clicks/taps
will still work.
NOTE that the above approach also takes care of weird edge cases where
you have a horizontal orientation on a device such as the iPad Pro that
is flipped in the horizontal position. Since clicks work anyway, these
edge cases are largely taken care of.
Anders Kaseorg
Mateusz Mandera
Alex Vandiver
Suyash Vardhan Mathur
Eeshan Garg
Abhijeet Prasad Bodas
PIG208
Priyansh Garg
Tim Abbott
Gaurav Pandey
Sahil Batra
akshatdalton
Ganesh Pawar
Alya Abbott