This adds a new client_capability that clients such as the mobile apps
can use to avoid unreasonable network bandwidth consumed sending
avatar URLs in organizations with 10,000s of users.
Clients don't strictly need this data, as they can always use the
/avatar/{user_id} endpoint to fetch the avatar if desired.
This will be more efficient especially for realms with
10,000+ users because the avatar URLs would increase the
payload size significantly and cost us more bandwidth.
Fixes#15287.
We need this field to avoid O(N) database operations
while fetching realm user data for clients with
`user_avatar_url_field_optional` flag enabled.
Part of #15287.
This extends get_accounts_for_email test by adding a deactivated
user and assert that get_accounts_for_email doesn't return any accounts
for that deactivated user.
Fixes#14807.
The zerver.models hack does not appear to be necessary now.
Meanwhile, get_wsgi_application has its own django.setup call, which
would overwrite the parts of our logging configuration pulled in by
zerver.models.
This fixes part of #15391; specifically, fixes it in production, but
not in development, where ‘manage.py runserver’ calls its own
django.setup and then imports various bits of our code before finding
zproject.wsgi.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
With #14378, we regressed back to the state of that
prior to 7e0ea61b00.
We fix this by getting our avatar bucket on
object initialization, and use the appropriate means
of gathering the network location for the urls.
Fixes#14484.
_setup_export_files modifies the zulip realm. We used to
call realm.refresh_from_db in tests after _setup_export_files was
called to make sure that the change is reflected. But sometimes
calling refresh_from_db was missed out here and there.
This commit makes calling refresh_from_db after _setup_export_files
unnecessary.
This commit adds message retention policy details in the subscription_type
text below the stream description.
We do not show any text when realm-level settings is set to forever and
stream-level is set to either forever or realm_default.
This commit adds frontend support for setting and updating message
retention days of a stream from stream settings.
Message retention days can be changed from stream privacy modal of the
stream and can be set from stream_creation_form while creating streams.
Only admins can create streams with message_retention_days value other
than realm_default.
This commit also contains relevant changes to docs.
This commits adds the code for live update of stream_post_policy in
subscription_type text in stream settings.
This is done by passing stream_data.stream_post_policy_values to the
template data, which were not passed previously and the if conditions
were not evaluated correctly.
This commit adds backend support for setting message_retention_days
while creating streams and updating it for an existing stream. We only
allow organization owners to set/update it for a stream.
'message_retention_days' field for a stream existed previously also, but
there was no way to set it while creating streams or update it for an
exisiting streams using any endpoint.
Because of a security release that required a migration, there
are two migrations numbered 0261. To avoid breaking existing
installs renumbering the migrations, we skipped migration 0261
when running tools/renumber-migrations.
Previously, we had implemented:
<span class="timestamp" data-timestamp="unix time">Original text</span>
The new syntax is:
<time timestamp="ISO 8601 string">Original text</time>
<span class="timestamp-error">Invalid time format: Original text</span>
Since python and JS interpretations of the ISO format are very
slightly different, we force both of them to drop milliseconds
and use 'Z' instead of '+00:00' to represent that the string is
in UTC. The resultant strings look like: 2011-04-11T10:20:30Z.
Fixes#15431.
There is apparently some way to have two instances
of `.emoji-popover-emoji-map`, although I can't
reproduce it.
This causes an `expectOne` check to fail fairly
deep in the stack.
Now we report it more directly.
This commit and a few previous ones mostly
address #15348 by trying to either a) not
depending on having a single instance of
the popover or b) making it more explicit
in cases where do expect that invariant.
Fixes#15348
This is just a pure refactor for now, but
we may want to modify this to more precisely
determine the active map (in case multiple
pickers are open for some reason).
This is clearly a better home for it, since message_scroll.js is the
only place that reads it, and also lets us provide a clearer name for
the functionality.
Since we are no longer using the "pointer" value sent in
page_params.pointer for anything, there's no value in continuing to
send it from the server to the client.
The remaining code in pointer.js is logic managing state for the
currently selected message.
Since the pointer is no longer used to set the browser's position, we
no longer need this complex code to send updates to the server during
the bankruptcy flow.
It's crazy that we need to do this; one would think that Electron apps
whose sole purpose is to be used with multiple team chat tools would
at least implement the standard desktop notification API correctly.
But it seems worth making this tactical change to prevent every
desktop notification throwing a traceback on those platforms, which if
nothing else results in a lot of error spam.
Fixes#15103.
This was broken when moving the code being called to another file.
This exception caused a pretty weird/nasty bug by interrupting the
message_fetch response handler before it finished updating the
fetch_status data strutures. The end result was that in views where
the "history limited notice" was displayed, local echo would be broken
a confusing notice would be displayed.
The term `parameter` is a better word than `argument` for data passed
to an API endpoint; this is why OpenAPI uses in their terminology.
Replace `argument` with `parameter` in the API docs to improve their
readability.
Fixes#15435.
In rare situations we would get tracebacks from
list_cursor on the line that I changed here. We
went the entire month of May without a traceback
here, and I can't reproduce the problem.
This is a pretty clear fix, though, and it will
hopefully lead to a more enlightening symptom.
The likely scenario here is that you use `q` to
navigate the stream list and then unsubscribe.
I tested that and couldn't get a traceback,
but I do think the traceback indicates some
possible issues.
The behavior I saw when I did this
appeared to be mostly harmless.
When I deleted a row (by unsubscribing), the code
seemed to effectively disable the cursor. It's
possible we should go to the next row or fully disable
the search.
I opened #15439 to follow up on this and other
cursor-related issues.
The stream_events tests were kinda messy, but
I mostly just consolidated a few sections of
code so that we didn't have to keep
re-stubbing the same functions.
For the actual code, I extracted add_sidebar_row
and then removed the unnecessarily complicated
jQuery trigger mechanisms.
This merges the `exports.get_search_result_legacy` and
`exports.get_search_result` function.
The key differences between the two code paths are as follows:
* We only want to generate suggestions for the queries which
the user is typing or can edit.
For the legacy version, suggestions are displayed for the
entire search string in the searchbox. (`all_operators`)
For the pills enabled version, suggestions are displayed
only for the input which hasn't been converted to pills.
(`query_operators`)
`all_operators` = `base_query_operators` + " " + `query_operators`.
trim is added at the end just to handle the legacy case
where we pass the `base_query` as ''.
* It is not possible to detect whether the user wants to
continue typing in the legacy version. However if the
the searchbox is still focused even after pill creation
we can assume the user still wants to continue typing.
To handle this we push an empty term as the `last` operator.
This is possible since the previous queries have been
completely entered as evident from it's generated pill.
* When using the legacy version, `search_operators` are
the same as `all_operators`, as mentioned in point 1.
In the pills enabled version we perform most of the
computations from the `query_operators`, but we do
require all `all_operators`, only for filtering the last
query's suggestion.
* And there is just one block unique to the legacy search
system. More details are mentioned in the comments of that
block.
We also refactor both the search suggestions node tests,
mainly to make them similar and easier to detect differences
when we switch over to the new version.
Hashir Sarwar
Karim Tayie
Anders Kaseorg
Tim Abbott
Wyatt Hoodes
Vishnu KS
Siddharth Varshney
sahil839
Clara Dantas
Rohitt Vashishtha
Steve Howell
Onur Temiz
Dinesh
orientor
Ryan Rehman