The work now being done in set_render_data used
to be done in render_more, which is called in
batches of 20 rows at a time. This work only
needs to be done once per narrow/view, so this
is a great performance improvement.
There were couple of problems in our handling of invalid or
incomplete URLs-
- The browser back button behavior breaks if someone enters
url with invalid group ID, incorrect group name.
- On typing group edit URLs with invalid group ID, an error
was raised and the URLs remained the same with just opening
the groups overlay with "Your" tab selected in left panel.
- On typing group edit URLs with incorrect right side tab or
without any right side tab, we showed "general" section, which
is fine, but the URL was still incorrect.
This commit fixes the above mentioned problems-
- URLs with invalid group IDs are now handled gracefully
with groups UI opening in the same way as before but the
url is updated to "#groups/your".
- We now update the right side tab to "general" if right
side tab is invalid or there is no right side tab.
- All the URL updates to fix invalid urls are done using
"history.replaceState" to make sure the browser back button
behaves as expected.
- All the code for checking the urls is done in hashchange.js
itself, so we can remove some code from change_state.
There were couple of problems in our handling of invalid or
incomplete URLs-
- The browser back button behavior breaks if someone enters
url with invalid or inaccessible stream ID, incorrect stream
name, "#streams/new" url without permission to create streams.
- On typing stream edit URLs with incorrect right side tab or
without any right side tab, we showed "general" section, which
is fine, but the URL was still incorrect.
This commit fixes the above mentioned problems-
- We now update the right side tab to "general" if right
side tab is invalid or there is no right side tab.
- All the URL updates to fix invalid urls are done using
"history.replaceState" to make sure the browser back button
behaves as expected.
- All the code for checking the urls is done in hashchange.js
itself, so we can remove some code from change_state.
The read receipts option, resides under the triple-dot message actions
menu. This made the process of viewing the read receipts take up
multiple steps, even via a keyboard-driven workflow.
Via this commit, now while focused on any message in a message feed,
and pressing `Shift` + `V`, efficiently brings up the read receipts
for that message.
Fixes part of #24716.
Co-Authored-by: SameepAher <sameepaher@gmail.com>
This was just debugging logic working around the fact that vdom.ts was
not originally implemented in TypeScript, and should not be polluting
our types.
This commit renames the realm-level setting
'signup_notifications_stream' to 'signup_announcements_stream'.
The new name reflects better what the setting does.
This commit renames the realm-level setting 'notifications_stream'
to 'new_stream_announcements_stream'.
The new name reflects better what the setting does.
Now whenever we initiate sending a message, we save / update its draft,
which is deleted on a successful send. Earlier, we did this only for
locally echoed messages. Hence a non locally echoed message's draft
would remain, if created in the timeframe between initiating send and
receiving the same message from the server, which can be significant
for slow connections.
For spectators, the chunk of page_params that originates from
do_events_register isn’t assigned until ui_init.js. That means the
TypeScript type of page_params is mostly a lie during module load
time: reading a parameter too early silently results in undefined
rather than the declared type, with unpredictable results later on.
We want to make such an early read into an immediate runtime error,
for both users and spectators consistently, and pave the way for
runtime validation of the page_params type. As a second step, split
out the subset of fields that pertain to the entire realm.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
For spectators, the chunk of page_params that originates from
do_events_register isn’t assigned until ui_init.js. That means the
TypeScript type of page_params is mostly a lie during module load
time: reading a parameter too early silently results in undefined
rather than the declared type, with unpredictable results later on.
We want to make such an early read into an immediate runtime error,
for both users and spectators consistently, and pave the way for
runtime validation of the page_params type. As a first step, split
out the subset of fields that pertain to the current user.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This commit extracts the edited notice computation logic of
edited_in_left_col, edited_alongside_sender and edited_status_message
to a new function - set_edited_notice_locations and calls it right
after the method merge_message_groups.
This is a preparatory commit, and is done since theses three variables
were previously wrongly initialized in the beginning for live messages
received.
We also introduce a modified flag to decide whether to display the
edited notice or not. This is useful since now we are computing
the edited notice values even when messages are not edited, and
hence we show the notices only when the modified flag is true.
The previous batch of improvements to this code path in
6562ea94e4 introduced a race bug where:
- You navigate to a narrowed view; Zulip renders cached data from
`all_messages_data` that we know is current, but
`fetch_status.has_found_newest` is initialized to `false`
nonetheless.
- The bottom is visible, triggering the check for whether the view
should be marked as read.
- `fetch_status.has_found_newest` is still `false`, and so we
incorrectly refuse to mark as read.
- We finish fetching data from the server, do the background rerender
for that (with no changes), but that doesn't trigger a re-check for
whether the bottom is visible.
There's several ways to address this, but most correct to me is to not
check fetch_status in this particular code path.
The same reasoning applies to the navigate.js call sites.
This should help reduce the risk of hitting rate limits when users
have a very large number of messages to fetch via this mechanism.
Inline the `messages` variable that was only used in one place while
we're touching this.
I was not able to reproduce obviously badly broken behavior from these
logic bugs, but after the renaming of message_viewport helpers in the
last few commits, it's clear that this logic was trying to check if
we're actually at the start/end of the possibly message feed, not just
the rendered portion, and doing so incorrectly.
So far, there were 2 separate turndown rules for code blocks; one for
general ones, and the other for Zulip message code blocks.
Now the filter rule has been generalised to handle both cases together.
As a side effect, the bug where partially copied Zulip code blocks
lost formatting on pasting has been fixed.
As discussed in the new comments, we had a bug where the
system-initiated animated scroll that happens when the compose box
opens as a result of narrowing would race with the internal
rerendering that occurs when the message_fetch request asking the
server for additional data returns.
The correct fix for this is just to open the compose box, if we're
going to do so, before setting the user's scroll position in the
narrowing/rendering process.
This ends up being a UI improvement (in that the compose box is
available for typing a bit earlier) as well as avoiding both the risk
of this race as well as the bad UX of adjusting the user's scroll
position multiple times as part of entering the view.
This does not address an as-yet-unknown bug wherein the animated
scroll that occurs when opening the compose box, when racing with a
background rerender, results in a bogus ending scroll position, though
it's easy to see how that might occur given that rerendering does
clear the DOM briefly.
Needed for typescript, because we want to preserve
types, so instead of mutating a message object,
we can instead calculate return these values
for a message object before it's created in full.
This commit also renames apply_markdown
to render, see this comment
https://github.com/zulip/zulip/pull/28652#discussion_r1470514780
Since notifications have limited space for the contents of a message,
a quote from a previous message, or elsewhere, can take up most of the
notification, leaving little room for the actual message, and reducing
the usefulness of the notification.
To fix this, we collapse blockquotes and "user said" paragraphs to make
space for the actual message.
From `get_current_nth_hash_section` `get_nth_hash_section()` is
extracted to allow for reuse for any hash. The indexing is also changed
to 0-based and negative indexing also works now.
This is a prep commit for the next, where `get_nth_hash_section()` will
be used in a new function with negative indexing.
Also, makes small updates to `next_plan_forms_support.html`.
Removes unneeded "btn" and "btn-default" classes, and updates
the placeholder text for the input as not marked for translation.
So far, when ordering typeahead suggestions, any query matches that did
not occur at the start of the target string were considered equally. So
for example, for the query "ok", "squared_ok" and "smoking" were
allotted equal priority, which does not make sense.
Now, matches from a word boundary (space, hyphen, underscore or slash)
are given priority (after exact matches and beginning matches), so that
in the above example, "squared_ok" is regarded as a better match than
"smoking".
Since recipients' sorting is complex, and needs word boundary matches
separately for decoupled sorting post triage, `triage_raw` is extracted
from `triage` to return the raw matches.
Fixes: #24127.
Though the correct mapping for Alt in Mac is Option, we had so far been
mapping it to Command, since for the 2 pre-existing shortcuts that used
Alt, Command was the only key that worked on Mac. However, these are
exceptions.
Now we map Alt to Option, and handle the 2 exceptions separately, using
data attributes on those shortcut elements.
This is a prep commit for introducing a hotkey that uses Alt / Option.
This will prevent any message we want to select after narrowing
from being offscreen entirely or partially.
Steps to reproduce the bug:
* `./manage.py populate_db -n 3000 --max-topics=2`
* Narrow to a stream and scroll high up.
* Align two recipient bars together with nothing between them.
* Click on the first recipient bar after keeping the selected message
on the second recipient bar.
* You will see that the selected message is not in view.
Fixes#26774.
We don't need the today argument anymore since every value passed to
it was a new Date() except the tests where we are using MockDate now.
All message view has `in-home` filter and is not empty.
If there are no terms provided, we set filter to display
all messages (including muted streams/topics).
We make it mandatory to provide terms to Filter.
For compose state variables `last_focused_compose_type_input` and
`message_type` that are assigned non-boolean values on user interaction,
`undefined` is the semantically better choice than `false` for
initialisation, to avoid incorrect boolean implications.
When pasting any code without any newlines, whether from a code block or
a code span, it will always be pasted as inline code.
That is, a line of code copied from within a code span will retain code
span formatting on paste, and a line of code copied from within a code
block will be pasted as inline code, instead of a 1 line code block.
Earlier, the tests for compose formatting were verbose, hard to read as
well as extend, and overly granular, without even having the ability to
test the final text selection or the cursor position.
Now, new test helpers, `init_textarea_state` and `get_textarea_state`,
have been added, enabling the tests to be more concise and readable,
while also being more powerful. A representative string alone now
describes the textarea state (the text and the selection / cursor),
making each test case as easy as defining the initial state as a string
and comparing the expected state post formatting with another string.
These new tests helped surface a couple bugs which have been fixed in
preceding commits.
Earlier, a new banner would be showed for each mention of the same
private stream in the compose box. This commit fixes that by checking
if the private stream warning banners already shown include the private
stream just mentioned, and if so, not showing a new banner.
This implementation is in line with the one for warnings for mentions of
users not subscribed to the current stream.
Fixes: #26914.
We previously allowed adding pills for inaccessible user by typing
the email manually. This commit updates it to handle inaccessible
users like we handle invalid emails.
Operand is a confusing name because this
data structure has an attribute *called*
operand. This commit renames references to
this structure to "term", short for "search
term".
For spectators, page_params.custom_profile_field_types has not been
initialized yet at module load time.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Some files already were using `noop` in place of `() => {}`.
It's both clearer what it means and is easier to type.
This updates all test files to fully use `noop`, and
adds a shared import from the test lib file.
The "user_group_name_dict" value for the old group name key
was not deleted and this led to a bug where the stream creation
UI was incorrectly showing about a user group already existing
with the old group name.
Fixes#28108.
evykassirer
Sahil Batra
Sayam Samal
Tim Abbott
Prakhar Pratyush
N-Shar-ma
Anders Kaseorg
Alex Vandiver
roanster007
Varun Singh
Aman Agrawal
Lauryn Menard
Shubham Padia
CIC4DA
Aditya Kumar Kasaudhan
Anuja Patil
Evy Kassirer
Vector73