We had a theory that get_user_id() errors were often due to race
conditions related to reloads, so we would only report missing
user ids if subsequent lookups failed 5 seconds later. It turns
out we still get the blueslip errors, and now we don't get
meaningful tracebacks. This change makes it so that errors
get reported immediately again.
"Local" datetimes are local to the server (or rather, are using
settings.TIME_ZONE), which in most cases is not what the recipient of the
message is expecting.
In this commit we just change the upload_avatar_image function to accept
two user_profiles acting_user_profile and target_user_profile. Basically
email param is dropped for a target_user_profile so that avatar's could
be moved lateron to user id based storage.
Now because it isn’t floating left it won’t slide down to a level below
the rest of the content when there are pixel rounding errors with
browser zoom.
Analytics database tables are getting big, and so we're likely moving to a
model where ~all stats are day stats, and we keep hourly stats only for the
last N days.
Also changed the name because:
* messages_sent_* suggests the counts (summed over subgroup) should be the
same as the other messages_sent stats, but they are different (these don't
include PMs).
* messages_sent_by_stream:is_bot:day is longer than 32 characters, the max
allowable length for a BaseCount.property.
Includes a database migration to remove the old stat from the analytics
tables.
If we get invalid events related to stream subscribers, we now
exit earlier to prevent ugly tracebacks. We may eventually
want to upgrade some of these warnings to errors, once we fix some
of our live-update bugs. In particular, we don't yet live-update
users when streams go from private to public, so if you add/remove
subscribers to a newly-public stream that a user still thinks is
private, they will not be able to handle the event through no
fault of the codepath that happens during the add/remove.
This currently only supports this in emoji reactions, not in actual
emoji in message bodies, but it's a great start for people who want a
text-only view.
Tweaked to update the text by tabbott.
Fixes#3169.
We have special code for closing a "compose preview", but
it should only apply to the enter key. Before this fix, we
would do the "enter" logic for other hotkeys like "j".
If you are typing a key like "q" in the compose box, there is no
need to check if the home tab is obscured, because the effect of
"q" is not limited by the message pane being opened.
This saves a bit of unnecessary computation when you type
non-arrow keys, which is especially important in the compose
box for characters that seem ordinary, like "j" and "q", but
which have mappings in some cases.
Standardizing the Zulip codebase to use UTC everywhere. Note that unlike
many recent commits in this line, this changes does result in a change in
behavior.
datetime.utcnow() is a timezone-naive datetime. The Django ORM interprets it
in the settings.TIME_ZONE timezone (e.g. 'America/New_York' in the
development server). We perhaps haven't noticed errors yet since with
'America/New_York' all it means is that emails are sent 5 hours early, or a
slightly different set of messages are included in the digest.
When you pass a naive datetime to the Django ORM, it uses settings.TIME_ZONE
for the time zone. In the development environment, both settings.TIME_ZONE
and datetime.now() use 'America/New_York', so there is no change in behavior
there. (fromtimestamp with no tz argument uses the same timezone as
datetime.now)
We are soon going to change settings.TIME_ZONE to UTC, so need to remove
naive datetimes from queries to the ORM.
This actually fixes previously broken behavior, since 'date' here gets
turned into the 'day' argument of seconds_active_during_day(day), where
tzinfo is set to UTC.
I noticed while reviewing #3807 that we still haven't fixed this;
because timestamps are primarily displayed in the message view, fixing
this is trivial.