When a user clicks outside the typeahead menu, inside the typing area,
the cursor position potentially changes, so `lookup` is called, which
considers the new cursor position and accordingly hides, continues
showing, or updates the typeahead menu.
This fixes the bug where even after clicking elsewhere, the old
typeahead menu continued showing and on making a selection, the text
was inserted at the wrong (new) position.
Fixes: #21302.
The upload spinners for all of the image upload widgets were in
wrong alignment due to the use of magic numbers to center them.
This commit replaces the above mentioned approach with the use
of flexbox to fix alignment issues across all the image upload widgets.
Due to differences in length of the words for different languages
there were alignment issues in the organization profile settings.
This commit uses flexbox to ensure that the alignment stays correct
for any changes in language/word length.
Fixes#21385
Previously, when a topic was edited (including being resolved), it
would become unmuted for any users who had muted it, which was
annoying.
While it's not possible to determine the user's intent completely,
this is clearly incorrect behavior in the `change_all` case, such as
resolving a topic.
The comments discuss some scenarios where we might want to enhance
this further, but this is the best we can do without large increases
in complexity.
Fixes#15210.
Co-authored-by: akshatdalton <akshat.dak@students.iiit.ac.in>
he possibility for it being null was likely an oversight -- it should
have been removed after the early migrations to backfill the field
when it was added.
We've confirmed there are no existing violations of this invariant in
Zulip Cloud.
Doing these in a loop may help us figure out whether the
flakes are somehow related to the initial conditions when
we run the test vs. some race that can happen later in the
loop.
I add the console statements mostly to facilitate debugging,
but they appear to actually reduce the problem, as the code
comments indicate.
We have a flake related to verifying that the app
prevents us from creating stream with duplicate names,
and my hypothesis is that it has to do with us not
waiting for the stream creation UI to fully appear. This flake
is probably a consequence of us recently making the stream
creation UI more like the stream editing UI, and thus
waiting for Desdemona to appear was giving us false
confidence that the page actually loaded.
I could be completely wrong about this solving the
flake, but the code change here is sensible regardless.
Specifically, this desupports:
android 4.4.3-4.4.4
baidu 7.12
ie 11
kaios 2.5
op_mini all
although we’ve already been blocking IE 11 since 3.0 (#14662).
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This is a natural follow-up to
93e8740218 - invitations sent by users
deactivated before the commit still need to be revoked, via a
migration.
The logic for finding the Confirmations to deactivated is based on
get_valid_invite_confirmations_generated_by_user in actions.py.
The ready method was deprecated in jQuery 3.0, because its behavior
has nothing to do with the selector; it always waits for the page to
become ready, not a specific element.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
It was broken by commit f5fbf5f0e0
“change_password: Migrate modal to dialog_widget” (#20193), because
the new_password input didn’t exist when we tried to install an event
handler for it.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
We now use narrow_state directly in pm_list and pm_list_data
tests, rather than mocking it with our `override*` helpers.
In some places I use an actual Filter() object, but
in places where the only testing concern is that the
active is narrow, I use a stub value.
We will continue to mock narrow_state in most places.
In addition to avoiding test-setup complications, we want
to avoid incidental line coverage on narrow_state that
only indirectly validates its behavior. Part of the
trickiness in avoiding narrow_state mocking is that
you often would have to introduce "real" Filter objects,
and the API for Filter objects is somewhat less than
ideal, and its wordiness can distract from the main
point of the tests.
Hopefully the changes here reflect the correct tradeoffs.
Co-authored-by: Steve Howell <showell@zulip.com>
Co-authored-by: Tim Abbott <tabbott@zulip.com>
This commit adds the backend functionality to
mark messages as unread through update_message_flags
with `unread` flag and `remove` operation.
We also manage incoming events in the webapp.
Tweaked by tabbott to simplify the implementation and add an API
feature level update to the documentation.
This commit was originally drafted by showell, and showell
also finalized the changes. Many thanks to Suyash here for
the main work here, which was to get all the tests and
documentation work moving forward.
Previously, we did not clear the resolved topic warning when reopening
the compose box after sending a emssage. This was a latent bug before
the previous commit, since the resolve topic was always displayed when
the compose box was pointed at a resolved topic before the previous
commit.
The resolve topic warning can feel like clutter in the event that the
compose box is empty (which often occurs when the user has no intent
to send a message), so we configure the validation logic to only
display the notice when the compose box is non-empty.
We take some care to minimize work the function is doing, beacuse it
is called on every keystroke in the compose box.
Fixes: #21155.
yogesh sirsat
N-Shar-ma
sayamsamal
Lauryn Menard
Tim Abbott
byshen-dev
Steve Howell
Anders Kaseorg
Mateusz Mandera
Alex Vandiver
Suyash Vardhan Mathur
Aman Agrawal