We have code that can automatically scroll an element into "view"
in its container. We use this for stream sidebar rows inside the
stream list.
Generally the stream sidebar rows are small enough to fit into
the container, and the prior algorithm worked correctly for that
scenario.
If you have lots of topics, however, and a short screen, the
algorithm was being too aggressive. For example, if the top
wasn't showing, it would scroll the top into view, but at the
cost of scrolling the bottom out of view.
This fix makes the general scrolling algorithm more tame.
Part of the user-facing problem is that the element we pass
into the scrolling code for the stream sidebar rows is bigger
than the part of the row that actually should be shown on
screen. Nevertheless, it makes sense here to make the general
algorithm more robust.
If you read a message, then got a topic edit for it, we were
adding the message to our data structure of unread stream/topic
messages.
Now we guard against this in unread.update_unread_topics. I
no longer expose an update() method in unread_topic_counter,
since we really want to do the unread check at a higher level
to keep other data structures consistent.
Category 'All' -> text 'Filter by category'; icon chevron right when
the dropdown is closed, icon chevron down when the dropdown is open
All other categories -> text CATEGORIES[state.category]; icon chevron down
A large portion of the diff for landing-page.js is due to refactoring the
contents of integrations_search into top level UI update functions.
State flows as follows: dispatch(action) -> render(state) -> update UI
Routes now use pushState instead of hashes.
On transition between categories scrolling position is fixed,
and on transition between catalog and integration sub-pages the page
scrolls to the top.
The lightbox "v" shortcut should not show a user's avatar,
so this limits the scope of images it can choose to ones inside
of the `.message_content` div.
This replaces the `startsWith` string prototype method with `indexOf`
because no versions of Internet Explorer support this feature, and
it really is not difficult to just use `indexOf` instead and check
whether the starting index of the full string is 0.
compute_placement utilizes the dimensions of the viewport, viewport location of
an element, and dimensions of an element to determine if a popover will fit
horizontally and/or veritically given its orientation. The default placement
is now viewport_center, which displays the popover, without an arrow, in the
fixed center of the viewport.
This should be particularly useful for hotspots on mobile or large popovers
that contain a lot of content. The property hotspot.location.popover can be
optionally set to fix the orientation of a popover (most likely to
VIEWPORT_CENTER).
This fixes the lightbox zoom issues that occurred on some browsers
due to the units of `deltaY` being in lines rathern than pixels,
making it incredibly slow to zoom.
This change simplifies how we mark all messages as read. It also
speeds up the backend by taking advantage of our partial index
for unread messages. We also use a new statsd indicator.
By the time we render messages, we will have set message.unread,
so we don't need to calculate it from flags.
We add a line to the local-echo path to make this explicit
in that code.
When we learn about updated message, a bunch of flag/boolean
fields concern us:
starred
mentioned
alerted
is_me_message
We now set booleans consistently with how we set new incoming
messages.
This code adds 'read' to message.flags and sets message.unread
to false.
It's not clear that the boolean message.unread is used in any
meaningful way, but we set it to false to avoid confusion. The
bankruptcy code was not doing this before.
Another quirk that existed before was that you could get two
'read' flags in a message when you declared bankruptcy. It's
also plausible that this could happen if you marked a message
as read via two different ways. It probably did not cause
user-facing bugs, but it would be confusing for troubleshooting.
Fixes#5032.
The new method borrows some code from the event loop
and unread_ops.mark_messages_as_read, and it is now
flexible about message_ids being marked as unread
even when there is no corresponding message in the
message store. For that scenario we still want to
update our data structures, which wasn't happening
before this change. (Generally, this was a non-issue
up until now, but it will become a bigger issue when
we start loading unread message ids from the server.)
This function allows us to see whether unread.js thinks a message
id is unread (as opposed to looking at the message itself). This
method is useful when we get notifications from the server that a
message has been read. In the future, we may not actually have
a local copy of an unread message, but we'll still know that it is
unread based on page_params. We'll want to update the data in that
case.
Going forward, we'll want to deprecate message.flags for most use
cases and just use the unread.js data structures to track unread
messages.
The prior implementation was needlessly complex. Both del() and
add() are cheap and idempotent.
With this change we no longer bother to delete a topic from a
dictionary when its last message is mark as read, since it doesn't
really help performance. We add a line to the tests to maintain
100% line coverage.
In a subsequent commit, we may have unread counts for
deactivated users. There is no reason to fail hard on these
scenarios; if there is no list item for a user_ids_string,
updating the unread count should be a noop.
It's not always clear whether user_ids are strings or integers, so
we explicitly convert them to integers for sorting when creating
keys for PMs.
To keep the tests passing, this commit removes some unneeded
defensive code in message_store.js that only applies to contrived
test input.
Main reasons:
* Shouldn't be hardcoding welcome bot
* compose_actions.cancel() was not closing the compose box, for some
reason. It was working fine before commit a few up from here ("tutorial:
Remove rest of tutorial."), but I think possibly due to the fact that one
had to click a button to exit the tutorial (that could be wrong, it was
hard to pinpoint why it was working before that commit and not after.)
This code should be going away anyway once #5816 is resolved.
* .screen is no longer being changed by other parts of the tutorial
* first_run_message we don't need, since we're guaranteed to have a message.
* Changing to #home and narrow.deactivate are not needed since we're
immediately narrowing to PM with welcome-bot.
These have been replaced by the initial stream messages and PMs.
The two pieces of information that exist here and not in the initial stream
messages are a link to /integrations, and a demonstration of image
uploading/pasting.
I think the current information presented is already a lot, though probably
it would be good to work in integrations somehow. Image pasting should just
be done in a separate Zulip that demonstrates the many different formatting
features.
When a user clicks on the `.player-container` node and the click
target is actually that (and not the YT video iFrame within), then
hide the lightbox – they likely mean to exit out of the lightbox.
This commit extract send_messages.js to clean up code related
to the following things:
* sending data to /json/report_send_time
* restarting the event loop if events don't arrive on time
The code related to /json/report changes the following ways:
* We track the state almost completely in the new
send_messages.js module, with other modules just
making one-line calls.
* We no longer send "displayed" times to the servers, since
we were kind of lying about them anyway.
* We now explicitly track the state of each single sent
message in its own object.
* We now look up data related to the messages by local_id,
instead of message_id. The problem with message_id was
that is was mutable. Now we use local_id, and we extend
the local_id concept to messages that don't get rendered
client side. We no longer need to react to the
'message_id_changed' event to change our hash key.
* The code used to live in many places:
* various big chunks were scattered among compose.js,
and those were all moved or reduced to one-line
calls into the new module
* echo.js continues to make basically one-line calls,
but it no longer calls compose.report_as_received(),
nor does it set the "start" time.
* message_util.js used to report received events, but
only when they finally got drawn in the home view;
this code is gone now
The code related to restarting the event loop if events don't arrive
changes as follows:
* The timer now gets set up from within
send_messages.message_state.report_server_ack,
where we can easily inspect the current state of the
possibly-still-in-flight message.
* The code to confirm that an event was received happens now
in server_events.js, rather than later, so that we don't
falsely blame the event loop for a downstream bug. (Plus
it's easier to just do it one place.)
This change removes a fair amount of code from our node tests. Some
of the removal is good stuff related to us completing killing off
unnecessary code. Other removals are more expediency-driven, and
we should make another sweep at ramping up our coverage on compose.js,
with possibly a little more mocking of the new `send_messages` code
layer, since it's now abstracted better.
There is also some minor cleanup to echo.resend_message() in this
commit.
See #5968 for a detailed breakdown of the changes.
Prior to this we were also performing highlighting inside HTML tags
which was wrong and causing weird behavior. Like, for example, if
someone added `emoji` as an alert word then any message containing
both emoji and alert word was rendered with a jumbo emoji.
Fixes: #4357.
This makes the canvas zoom and pan feature cross-browser compatible in
a few ways:
1. Replace deprecated `mousewheel` event in favor of the similar and
cross-browser event `wheel`.
2. Create approximate substitute for `e.movementX` and `e.movementY`
values that are missing in Safari.
This completes the major endpoint migrations to eliminate legacy API
endpoints from Zulip.
There's a few other things that will happen naturally, so I believe
this fixes#611.
A realm filter should match only after the start of a line, whitespace
or opening delimiters. But markdown was not configured to respect those
rules which was causing some weird rendering behavior. This commit fixes
the regex used for matching realm filters. On the backend we are using
regex with negative lookbehind to perform matches but since javascript
regex don't support lookbehind we are using a workaround on the frontend
using `contains_backend_only_syntax()` function which detects if a realm
filter can be rendered correctly by backend only and if so it stops the
message from getting echoed locally.
Fixes: #5154.
This class is mostly a thin layer over the dictionary, but it
consolidates all the logic to create lookup keys, which have
to follow the convention of being comma-separated, numerically
sorted user_ids.
interface_type select menu will be used to choose the interface
for outgoing webhooks. It will be displayed only when the selected
bot type is OUTGOING WEBHOOK type. The default value is GENERIC
interface type (1).
It appears that a regression introduced in
3f60074c33 caused undefined to be passed
as the subject to the recent_senders library much more often; this
fixing that, and makes the library handle such cases reasonably
without an exception regardless.
This was causing a huge number of "Tried to call a Dict method with an
undefined key." exceptions.
We now call topic_data.add_message() and
topic_data.remove_message() when we get info about
incoming messages. The old way of passing in a boolean
made the calling code hard to read and added unncessary
conditional logic to the codepath.
We also have vague plans to change how we handle
removing topics, since increment/decrement logic is now
kind of fragile, so making the "remove" path more explicit
prepares us to something smarter in the future, like just
figure out when the last topic has been removed by calling
a filter function or something outside of topic_data.js.
Another thing to note here is that the code changed here
in echo.js is dead code, since we've disabled
message editing for locally edited messages. I considered
removing this code in a preparatory commit, but there's
other PR activity related to local echo that I don't want
to conflict with.
One nice aspect of removing process_message() is that
the new topic_data.js module does not refer to the legacy
field "subject" any more, nor do its node tests.
This commit introduces a per-stream topic_history class
inside of topic_data.js to better encapsulate how we store topic
history.
To the callers, nothing changes here. (Some of our non-black-box
node tests change their way of setting up data, though, since the
internal data structures are different.)
The new class has the following improvements:
* We use message_id instead of timestamp as our sorting key.
(We could have done this in a prep commit, but it wouldn't
have made the diff much cleaner here.)
* We use a dictionary instead of a sorted list to store the
data, so that writes are O(1) instead of O(NlogN). Reads
now do sorts, so they're O(NlogN) instead of O(N), but reads
are fairly infrequent. (The main goal here isn't actually
performance, but instead it just simplifies the
implementation.)
* We isolate `topic_history` from the format of the messages.
This prepares us for upcoming changes where updates to the
data structure may come from topic history queries as well
as messages.
* We split out the message-add path from the message-remove
path. This prepares us to eventually get rid of the "count"
mechanism that is kind of fragile and which has to be
bypassed for historical topics.
The var in question is indexed by stream_id, so stream_dict seems
like a good name for it. We want to distinguish per-stream data
structures from the stream-level entry points.
This new module tracks the recent topic names for any given
stream.
The code was pulled over almost verbatim from stream_data.js,
with minor renames to the function names.
We introduced a minor one-line function called stream_has_topics.
We now have all of our callers into recent_topics code just
receive a list of topic names from get_recent_topic_names().
This is more encapsulated than handing off tiny little
structures to the three callers, two of whom immediately
mapped the objects to names, and one of whom needlessly
used the now defunct name canon_subject field.
The consolidation here removes some "subject" references, and
now all lookup are by stream id, not stream name.
The diff here is a bit daunting, but it's mostly simplification
of tests and calling code. Two of the callers now need to look
up stream ids, but they are otherwise streamlined.
The main change here is to stream_data.js, and we replace the
`canon_subject` and `subject` fields with `name`.
This is an issue where in jQuery v3 the result of outerHeight on a node
that doesn’t exist is now “undefined” rather than “null”, which means
it will no longer cast to a Number but rather NaN. For this, we create
the safeOuterHeight and safeOuterWidth functions to safely return a
result (or 0).
This is a better solution than manually going to each instance and
ORing it with 0 for type safety.
https://stackoverflow.com/questions/41454285/jquery-outerheight-returns-
undefined-instead-of-null
This removes the `no-new` rule which is relatively detrimental to
code cleanliness in our codebase because third-party libraries may
utilize data structures that don't fly well with our linting rules.
This also fixes abstractions that were created due to the limitations
and impositions of this lint rule.
This changes a typo where a function was attempting to execute the
scope of the parent's "this" by using `function () {}.on()`, rather
than using the `Function.prototype.bind` built-in.
The ES6 backticks in /static/js/portico/landing-page.js were causing
the server to trip up. This was fixed by switching from JS string
interpolation to string concatenation (ES5).
Modified timerender.js absolute_time() to include the year
in the returned string when the supplied timestamp is in
an older year. This included adding an optional second
argument to specify the current date to facilitate unit
tests.
Fixes#5737.
This commit specifically addresses the issue when in preview mode,
while "enter sends" is enabled. Previously the messages were just
sent, now they must pass validation.
Fixes#5574.
This allows us to reliably parse the error in code, rather than
attempt to parse the error text. Because the error text gets
translated into the user's language, this error-handling path
wasn't functioning at all for users using Zulip in any of the
seven non-English languages for which we had a translation for
this string.
Together with 709c3b50f which fixed a similar issue in a
different error-handling path, this fixes#5598.
This fixes the original issue that #5598 was the root cause of; when
the user returns to a Zulip browser tab after they've been idle past
the timeout (10 min, per IDLE_EVENT_QUEUE_TIMEOUT_SECS), we now
correctly reload the page even if they're using Zulip in German or
another non-English language where we have a translation for the
relevant error message.
This function no longer sets properties to false, so the supported
way of doing this is to instead use prop(foo, false). Some tests
had to be fixed to accommodate this.
In case the user was not allowed to upload an emoji, we were displaying
two different but sematically same tips. This commit merges them and
also updates `update_custom_emoji_ui()` function in settings_emoji.js
to live update tooltip.
Our code to edit messages that were echoed locally but failed
by the server was broken. We just disable it for now.
We have opened #5841 to try to restore this functionality.
Our logic for editing failed messages is broken in various ways,
so we are removing the codepath for editing for now. We will
try to restore these features as part of #5841.
Because of local echo, message ids can change in message rows.
Having reactions use markup to indicate their message id just
creates more moving parts, since we would need to handle
message_id_changed events.
Now our handlers just call row.get_message_id() as needed.
We no longer do the message_store piece of reifying ids
via a trigger. We now make an explicit call to an
ordinary function.
This has several benefits:
- no more initialize() function
- no more scary comments about garbage collection
- the function has a real name now
- the function is less indented
- we can easily see when the message_store step happens
- simpler node tests
- simpler tracebacks (no jQuery cruft)
If you use the escape key to close a message edit, we need
to blur out the text fields. Otherwise, hotkeys.js thinks
we are still editing the text. This bug would disable the
use of things like arrow keys until the user subsequently
focused another field.
We probably eventually want hotkeys.js to be smarter about
ignoring hidden fields that still have the focus, but there's
also no reason not to blur the fields here, and this is a more
local, less risky fix.
This commit renames possibly_notify_new_messages_outside_viewport()
to the more concise name notify_local_mixes().
We really only need to call this function in one place, so we
have the caller check the `local_id` condition. We can eventually
upstream this code even further so that it's completely
obvious that it's only ever called from the local-echo codepath.
We were calling maybe_add_narrowed_messages() in a place
where local_id is guaranteed to be undefined, since
we always set local_id to undefined when
can_apply_locally() fails.
In turn maybe_add_narrowed_messages() was calling
possibly_notify_new_messages_outside_viewport(), which
requires a local_id to do anything meaningful.
This removes all the associated dead code--passing in
a parameter that we know always was undefined and
calling a function that we know always would no-op.
Not only does this simplify the code a bit, but it avoids
us stepping on the toes of the alternative code path that
deals with non-locally-echoed messages.
This commit early-exits before our loop when local_id is none,
and it tries to more clearly indicate that the callers will
generally be just calling this with messages sent on the
local-echo path.
We now make it so that get_next_local_id() only returns up
to 5 local ids relative to any given max id.
For example, if your pointer is at message 999, we'd give out
999.01, 999.02, ..., 999.05.
We also avoid giving out the same local id twice. This prevents
a bug where if you had aborted a previously failed locally echoed
message, a subsequent local echo would get into a funny state.
In order to to prevent false alarms on using the same local id
twice, we call get_next_local_id() later in the try_deliver_locally()
function.
We had code that would try to re-render locally echoed messages
that were rendered right before a browser restart. This code
has gotten buggy over time, so we are removing it for now.
We will try to re-solve the problem as part of #5825, but
possibly with a different strategy.
This allow the webbpack dev server to properly reload JavaScript modules
while running in dev without restarting the server. We need to connect
to webpack-dev-server directly because SockJS doesn't support more than
one connection on the same host/port.
We eliminate `.get(0)` calls in buld_stream_list.
The easy case is that we stop building jQuery objects
for the splitters only to pull out the DOM immediately.
The more subtle case is that we also don't do `.get(0)` calls
to get DOM out of our individual list items. By passing
in full jQuery objects to `append()`, we should prevent ourself
from orphaning the old objects, which may in the future have
things like tooltip logic attached to them.
This either removes aria-hidden=true assignments from buttons with
text, or adds a span to only hide the 'x' symbol rather than the
button for closing buttons.
This new setting controls whether or not users are allowed to see the
edit history in a Zulip organization. It controls access through 2
key mechanisms:
* For long-ago edited messages, get_messages removes the edit history
content from messages it sends to clients.
* For newly edited messages, clients are responsible for checking the
setting and not saving the edit history data. Since the webapp was
the only client displaying it before this change, this just required
some changes in message_events.js.
Significantly modified by tabbott to fix some logic bugs and add a
test.
I deleted a test case that involved a highlighted stream, but
the query was empty. This produces kind of a weird result with
typeahead_helper.highlight_with_escaping, but this function already
has coverage in node_tests/typeahead_helper.js, so the check here
was essentially redundant anyway. Specifically, the highlighter
wraps every character individually with <strong>, and looks really
messy in html.
Fixes#3496. This was just a simple regex addition to filter
to accept the format `operand:"foo bar"` as a token. Also,
it will now accept an additional space after the separating colon.
I pushed a bunch of commits that attempted to introduce
the concept of `client_message_id` into our server, as
part of cleaning up our codepaths related to messages you
sent (both for the locally echoed case and for the host
case).
When we deployed this, we had some strange failures involving
double-echoed messages and issues advancing the pointer that appeared
related to #5779. We didn't get to the bottom of exactly why the PR
caused havoc, but I decided there was a cleaner approach, anyway.
The old translation copies in localStorage were not being removed
when they were no longer needed, so we can free up the storage
by deleting them.
This was accidentally not merged months ago when originally
implemented, but it was written to fix#4443 and in fact does so.
This change has us tracking messages as soon as we start
sending the message to the server. The next step is to
reconfigure the timeouts a bit to deal with the server not
responding.
We now use a client-side message id to track the state of our
sent messages. This sets up future commits to start tracking
state earlier in the message's life cycle.
It also avoids ugly reify logic where we capture an event to
update our data structure to key on the server's message id
instead of the local id. That eliminates the node test as well.
Another node test gets deleted here, just because it's not
worth the trouble with upcoming refactorings.
This mostly sets the stage for a subsequent commit to start
using client_message_id as the key into sent_messages.
It has the nice side effect of making it more explicit that
certain things should always happen when transmit_message()
succeeds.
This commit does regress our node test coverage a bit.
This commit starts to decouple client_message_id from local_id.
We don't really take advantage of the decoupling in this
commit--in fact, it's a bit of a pain at first. But this should
be a fully working checkpoint commit.
This is mostly straightforward moving of code out of compose.js.
The code that was moved currently supports sending time
reports for sent messages, but we intend to grow out the new
module to track more state about sent messages.
The following function names in this commit are new, but their
code was basically pulled over verbatim:
process_success (was process_send_time)
set_timer_for_restarting_event_loop
clear
initialize
All the code in the new module is covered by previous tests that
had been written for compose.js. This commit only modifies
a few things to keep those tests.
The new module has 100% node coverage, so we updated `enforce_fully_covered`.
We are deprecating local_id/local_message_id on the Python server.
Instead of the server knowing about the client's implementation of
local id, with the message id = 9999.01 scheme, we just send the
server an opaque id to send back to us.
This commit changes the name from local_id -> client_message_id,
but it doesn't change the actual values passed yet.
The goal for client_key in future commits will be to:
* Have it for all messages, not just locally rendered messages
* Not have it overlap with server-side message ids.
The history behind local_id having numbers like 9999.01 is that
they are actually interim message ids and the numerical value is
used for rendering the message list when we do client-side rendering.
Prior to this commit, 7 megabytes of images (through 253 individual requests)
were heavily slowing down the initial load. With this commit, we load only the
logos (60 or so images).
Documentation and images for the individual integration sub-pages is requested
separately using the /integrations/doc/ endpoint, which returns HTML.
This redesigns the /help/ page sets to be a single page app that uses
history.pushState to work the same as the old app.
The big new feature is that now we have the index in a nicely designed
left sidebar.
`add_emoji_by_admins_only` backend setting is represented by page_param's
`realm_add_emoji_by_admins_only` attribute. When this setting was changed
we were wrongly updating the `add_emoji_by_admins_only` attribute which
doesn't exist.
Prior to this, when the setting for controlling whether can admins only
upload an emoji was set to true, we were not displaying upload emoji form
even for admins and as a result they were locked out.
The piece of code is dead since there can be no instance where
email === ''. This is ensured by util.extract_pm_recipients
by filtering for empty strings in the pm_recipients list.
This piece was dead because exports.send_times_data[message_id]
cannot be undefined since the only place this function is called
from is exports.report_as_received() and that function has a call
to mark_end_to_end_receive_time() before a call is made to the
function in question for dead code. The function call to
mark_end_to_end_receive_time results in
exports.send_times_data[message_id] = {} if this was not defined
already. So there can be no instance where we end up the code
being removed.
In typeahead_helper.js, added a compare function to first sort by
subscription, then by pm partners and lastly based on recency in the
current topic. Altered function sort_for_at_mention to take topic data
and sort using the above function. Also altered node tests for
typeahead_helper.js to test for the above added functionality.
Fixes: #4249
There is no reason to render the template for compose mention
warnings if the user is already in the widget.
This commit also restructures the unit test significantly to more
carefully exercise each case, particularly in regard to when
templates get rendered.
Creating a new bot (by filling out the bots related fields and clicking
"Create bot" button) changes tab from "Add a new bot" to "Active bots".
This is done to make users know/confirm that the bot has been created and
the user can view it in this tab.
Fixes: #5731
Our current workflow for creating a new stream allows the user to
invite as many other users as they like but since there can be
mistakes in doing so, we now open a modal with a warning if the
number of invites are more than 100 just to confirm that user indeed
wanted to do this.
Fixes: #1663.
These are some strings I spotted in English when playing around a bit
with the UI set to German, where our translations are near complete.
It'd be great to have a more systematic way of spotting this kind of
omission. Probably a fairly simple linter could catch a lot of cases.
Without this, the "Since you were last here" text got rendered in
English for me every time when I tested in German, both in dev
and on chat.zulip.org.
This brings us to 9 places we invoke `ensure_i18n`. That seems
like a sign that there may well be more places we're still
missing, and that we should probably find a more systematic way
to make sure all our frontend UI rendering waits for translation
to be ready. Anyway, for now, fix this one.
This system hasn't been in active use for several years, and had some
problems with it's design. So it makes sense to just remove it to declutter
the codebase.
Fixes#5655.
Fixes#5612. What this specifically does is that if you are
typing a group PM, this logic iterates through the possible
search suggestions for the next autocomplete. If that suggestion
contains a group PM that already exists, then prioritize it with
the most recent one on top.
Steve Howell
Jack Zhang
Brock Whittaker
Greg Price
Vaida Plankyte
Tim Abbott
Steve Howell
Rishi Gupta
Joshua Pan
Harshit Bansal
vaibhav
Umair Khan
Vishnu Ks
digi0ps
David Coleman
hollywoodno
Sarah
Cory Lynch
Shubham Gupta
Cynthia Lin
Pweaver (Paul Weaver)
Durga Akhil Mundroy
Aditya Bansal
Abhijeet Kaur