Previously topics weren't being highlighted at all and messages had their
highlighting persist across different narrows (because we were only checking
whether the message object had a match_content property, not whether it should
currently be used).
(imported from commit 44c91c6d5799dcdf765e19e1a17bd727ce80c918)
When navigating away get_events will fail after we delete the event
queue before leaving the page. In that case we try to reload the page to
correct the problem. This happens before the page navigation is
complete, and then we reload zulip keeping the user captive. This was
only observed on on Firefox.
(imported from commit e001172e87a9f2ab7cf07a477e46b9d87752ac04)
Now that we no longer use tables for our message list, we can
more logically group messages together.
(imported from commit 9923a092f91a45fe3ef06f2f00e23e4e3fb62a37)
This experiment has been disabled for everyone for a while: if we
bring something like this back, it is not likely to be exactly the same,
and will be different enough to require a different implementation.
As it is, the summarization code was making a few code paths (rendering
especially) more complex, and is worth removing for simplicity's sake.
(imported from commit 6ac8cdc9f7077a5a1da01ab4268aba3db0bc43f8)
If you have a lot more individual users in your realm than group
PMs in your recent history, we were squeezing out the Group PMS.
We now max out the ratio for any sub-section in the right sidebar,
as well as bumping up the min.
(imported from commit f7c44367f2a518d27406993cb6358cc96d1aae92)
This changes Filter.describe and Filter.operator_to_prefix
to handle negated terms correctly.
(imported from commit 673c0d3a5a77784e95772c14e12534ad2daecda2)
Extract 'verb' in Filter.operator_to_prefix() and
Filter.describe(). This doesn't change any functionality;
it is designed to make a subsequent change have a less
noisy diff.
(imported from commit e9c4b6edc498d88ec1783ccdba079d980def9438)
The methods describe() and operator_to_prefix() are
now in filter.js. They seem like they eventually
may be useful outside the scope of search suggestions,
and they seem to go hand in hand with other Filter
responsibilities.
(imported from commit 55f14c3ee848febb2e7c8c6a27afd690569b4e08)
Also:
* Change fixed element positioning and fix bugs
* Move settings dropdown back to the right and add left padding to left sidebar
(imported from commit fcf903b59617687f94618a01ce7544b69f408130)
Remove the options to narrow by topic/person from the menu,
because there are better ways to do this in the UI, and
remove the time travel option, because the "Link to this
conversation" achieves mostly the same effect.
(imported from commit b7e0cfe64c0760e5a7bf7a8c9c05ed1a5b747300)
This reverts commit 462a3eb5e6b83f9d8091b83e3f8dc458236938ed.
We're reverting this to see if it is the cause of our recent CSS
performance issues.
(imported from commit 6a0b041cfcb6770bbfda0d354444bad2d64459ab)
For the Filter helper functions above, we generally want to
ignore negated search terms, since their existence should
really only impact filter predicates and nothing else on the
JS side. The exception is search, where even the existence
of a negated search needs to be noted to know that we can't
apply a filter locally.
(imported from commit 8bbb410a85fefed549d359e4c779a134ad830c11)
For negated search terms, we weren't explicitly setting
"negated" to false when callers left it undefined, which was
mostly fine, since undefined is falsey, but it is better to
define it explicitly for debugging/testing purposes.
(imported from commit 68a2790b510d17caed8ca11c38188545d1dcc347)
Behind a feature flag you can now do searches like this:
-pm-with:othello@example.com is:private
The "-" in front of "pm-with" tells us to exclude messages
with Othello from our search. We support "-" in front of
all operators, although the behavior for "-search:" and
and "-near:" doesn't really change in this commit.
Note that the filtering out of "negated" predicates only
happens on the client side in this commit. On the server
side we ignore negated predicates and send back a superset
of the results.
(imported from commit 6cdeaf32f2d493fbbb838630f0da3da880b1ca18)
IE sometimes returns the pathname without a leading slash. Also
location.origin is not supported and must be build manually.
(imported from commit fb64478aeaac0f17d31021b7c370ff56781b48d1)
This this removed one forced relayout of the page on unnarrow. This
saves about 100ms for me.
(imported from commit 0755f425abbe3d99b8a99765549a5bbf3c620b9a)
Previously, we saved the current_msg_list selected id and then
restored it as the home_msg_list selected id, which could result in
the home view loading to the wrong place.
This takes some already bad code and makes it even more in need of
refactoring, but it does fix a pressing bug. We should definitely
refactor both:
* the top of narrow.js
* the save/restore code in reload.js
after this, though.
(imported from commit bb2040219e4f545ba90bb04a696996cec2831484)
This reverts commit 64aced74012101f3bdbd3a4e6066b46ad8e1f4ea,
which was always intended as temporary code until we upgraded
the back end to support dictionary-style narrow operators.
(imported from commit b8d3a19f3aad3d9d6a26b9dcc07f502c55b18edd)
The filter_term() function was supporting the transition
from using tuples for search terms to using dictionaries,
but now all of the JS code should be dictionary-compatible.
(We had already abandoned the tuples safety net on staging,
and a couple days of use have given me confidence we can
pull the shim code.)
The one side effect this change has is that search terms will be
initialized to {} instead of []. This distinction matters
when it comes to calling JSON.stringify on the search terms.
(imported from commit 1fbe11011d8953dbea28c0657cbf88384d343e00)
The channel module now keeps track pending ajax requests and has an
abort_all function to angle all pending requests.
(imported from commit 4e78ab24d2295bd67de5633e3a200dfa489825b1)
When we typed "stream:" into the search bar, the empty operand
triggered an error in the Dict class for an undefined key, because
we were using opts[0] as a "defensive" workaround to opts.operand,
but opts.operand of '' is more correct than opts[0] being undefined.
Now we only fall back to opts[0] whe opts.operand is undefined, and
we emit a blueslip error when that happens.
(imported from commit 88a196d3bc3d67689c36bc036f378da744c652f9)
Have the server send down the stream's id for removal
events, and have the client use that id to look up the
stream in its internal data structures. This sets the
stage for eventually just sending the stream id (and not
the stream name) down to clients, once all our clients
are ready to use the stream id.
(imported from commit 922516c98fb79ffad8ae7da0396646663ca54fd0)
Splitting out subs.mark_sub_unsubscribed gives the calling
code flexibility to track down the "sub" as it sees fit.
(imported from commit 4052f5a2a0f6fd58a2177f2b5961ce72f052cf94)
Before this change, we were using sequentially generated ids
on the client side to identify streams. Now we just use
the ids from the server. The goal here is to reduce the
confusion of having two different ids attached to a stream.
Also, not that it matters a ton, but this also means that
the browser basically has an immutable id for each stream
that is future-proof to reloads, multiple create_sub calls, etc.
It also a bit easier to grep for ".stream_id" than ".id".
(imported from commit 057f9e50dfee127edfe3facd52da93108241666a)
We have shim code that makes our internal narrow operators
support both a tuple interface and an object interface. We
are removing the shim on staging to help expose any dark
corners of the code that still rely on operators being
represented as tuples.
(imported from commit f9d101dbb7f49a4abec14806734b9c86bd93c4e1)
Our overall guideline is the type names for events are singular, and the list of
events of that type are plural. 'subscriptions' was not following this guideline
and (potentially as a result) had a bug where it was impossible for clients to explicitly
subscribe to subscription change events properly.
(imported from commit 7b3162141fd673746e0489199966c29ea32ee876)
Chrome was showing a memory leak after many auto-reloads. Emptying the
the collections and removing the event listeners reduces the severity.
Before this change 40 reloads would would grow to about 140MB, now it
stays around 50MB.
(imported from commit 55fbeff9bdd0363bb95929f2981a2de238ff35d8)
This is yet another change related to phasing out the
[operator, operand] tuple data structure for representing
terms in a narrow.
(imported from commit 508e58fc4eebae8a24a8ae59919ba5d94fc66850)
The JS code can now call stream_data.get_sub_by_id() to get
a sub from a stream_id. Subs have stream_id due to a prior commit,
and we keep track of the mapping in stream_data's subs_by_stream_id
variable.
(imported from commit 409e13d6d2e79d909441a66c85ee651529d15cd2)
The tutorial introduces "engineering" messages that might not
be in the user's normal subscription, and they would get a gray
border if we did not override the stream color. Before this change,
we accomplished this by overriding the core data structure in
stream_data.js. Now we are a bit more future-proof; we only
override stream_color.default_color.
(imported from commit 0d0845b72f766912679f5aa7641ae9a60fdbb4ce)
This does have a small functional change of using use_closest even for
the current_msg_list.last().id case, but that's harmless; and it does
reduce a lot of code duplication and confusing logic here.
(imported from commit 7c4ecaa197120cc6d5c05ce4887f33c7d94a9c59)
The leading theories this will test are:
1) MessageList._items becomes unsorted and the binary search starts
returning the wrong index from selected_idx.
2) MessageListView.render is not rendering the message or it is not
being inserted into the cache.
(imported from commit 926b7d7bf9df338320a8cba6241038176ce4b47d)
This sets us up to redraw the window title without having to
pass in the unread count, which will be useful for realm name
changes. The redraw_title() function is only responsible for
rendering stuff, whereas update_title_count() handles
the details of caching the count.
(imported from commit 67cfb7a273e2a61720ce2bd8d91d73fe1089c813)
Add try/catch blocks to get_updates_success and send a blueslip error on
errors we catch. This will let get_updates_success return successfully
so that the next call to get_updates will start immediately.
(imported from commit 44d8b85d9d8e930a5552a5fbf4af1d0e5e8c07e8)
There are 2 uses of all_msg_list previous to this commit:
(1) The contiguous block of messages to be used for constructing the
initial state of narrows.
(2) A way to look up an arbitrary message by ID that may or may not be
in the home view to check if we have it locally.
We eliminate all applications of use case (2), replacing them with
queries on message_store, since they are in fact wrong -- any messages
that are outside the contiguous time period of all_msg_list would not
appear in all_msg_list and thus would incorrectly not be returned.
(imported from commit e2e2efe919242331bbc44abc6c50b16e3ab1336e)
This implementation is somewhat hackish in large part because I think
we're going to be wanting to redo the get_old_messages API somewhat
soon, and this may naturally become a lot cleaner as a result, but
this isn't a lot of code and fixes#2235 part (A) and substantially
mitigates #1510.
(imported from commit 47a2160a44befa9d83190c5cc95b90e92cc5b4cc)
When users select messages starting with stream header we should format
the text as using the custom copy handler.
(imported from commit 4f9851939484d046619c3e53c71c47ad64797670)
I apparently screwed up when backing up the process_loaded_for_unread
move in a way that just lost the function.
(imported from commit 91dfcf1abc85d439274cb8b0be380e9230942ebb)
This is no longer required, now that we're no longer scrolling the
main body of the page.
(imported from commit c2aa0d403c8fd0679b3110fe8e7684d46a7557fa)
Both pointer and last are fields that were removed when we changed the
protocol to use events like a year ago.
This conveniently allows us to stop exporting get_updates_params to
other modules.
(imported from commit 13651ee240ce6e1bafb241b185e5ef653c348cdf)
Previously, we waited until we'd fully loaded the home view to start
calling get_events, which:
(A) didn't make sense non-message related events
(B) resulted in our throwing "get_updates not started" errors when
users sent a message before the home view was fully loaded.
This commit only fixes problem (B); to fix problem (A) we will need to
change the event capture code to only capture some events.
(imported from commit 2aacd6bdfd1e7d7707e3a724f1fde7da4a556538)
Note that this makes the event fire after the home view has messages
loaded all the way to current; previously it only waited for the first
batch of messages to load.
(imported from commit a5c408ea0136c761f4cc77e8be4977b13acc6d77)
By using _.all() instead of a for loop, we avoid a tiny bit of
confusing between "break" being for a switch statement and being for
the loop.
(imported from commit cd6e7ff788b50f4dadce93e7f0efcb381bc59270)
Even in the case where we are running embedded narrows,
we want narrow.operators() to return new-style operators
that you get with Filter objects, so that callees do
not need to support the legacy tuple format.
(imported from commit a6649881a926a2304e9f7cc8ca53b406e2c8828e)
If you do a search like id:5 topic:foo and message #5
does not have the topic "foo", we now return zero results.
(imported from commit 8121fac1dbd79024c51af1f310d831dab9242e36)
Have narrow.activate() user filter.operators() to produce
operators that have operand/operator fields, so that its callees
don't need to be backward compatible with the tuple format.
(imported from commit e408e33074d1be2d112bb3cdc081ec3616c908ee)
Now that we are not scrolling the document we can remove the workaround
to prevent browsers from scrolling on hash changes.
(imported from commit 67fdaa4980d4d54d80ca9c259bbee567b8eeb917)
Change the offset to hold of the selected message as an offset from the
top of the screen. Then use the current offset and scrollTop to compute
the new scrollTop.
(imported from commit 718e95d3435c0f84cbb7663a9bb2bc2789314203)
By having Filter.canonicalize_tuple() call filter_term(),
we make it so that Filter objects get operator/operand
fields in their terms when we initialize this.
This mostly caused test breakage for tests that were doing
assert.deepEqual; now we just check to make sure that the
field we need are there.
(imported from commit 63b2516dc72edeb11e76a1fa4442570b9c605baa)
Consumers of Filter.parse() can now reference
search term parts like so: term.operand, term.operator
(Legacy code can still use term[0] and term[1].)
(imported from commit 06d0da65f13f1eb7e3ba8eac0e69448aab2735ab)
This got broken in the local echo refactoring -- we accidentally
switched the update_starred function to be a toggle, which meant that
the message would be marked as starred and then unmarked as starred
moments later, due to the two update_starred calls for the initiating
browser (one from when the click happens, the second from when the
event returns from the server).
(imported from commit 8f83729fe5477cf052124c1c56ed9189b85b885c)
Previously, while you'd get the event saying you'd been knighted,
which would make the Administration tab visible, clicking on the tab
would error out because the admin page HTML was never sent over on
page load (since you weren't an admin at that point).
(imported from commit 90ad351533515bebece630d67baf4b142d320754)
Clients can receive description updates for streams which are not in
stream_data. We should ignore the updates and warn that this has
occurred.
(imported from commit 3cd5a98a133012ae27ea0f43b03cc9d8e459e49c)
Leo Franchi
Allen Rabinovich
Jason Michalski
Luke Faraone
Zev Benjamin
acrefoot
Steve Howell
Jessica McKellar
Tim Abbott
Waseem Daher