It's always been the case that in production, Tornado dumps all the
event queues when shut down so that they can be reloaded by the
replacement Tornado process. This never worked in development because
the codepath for auto-reload didn't go through either a signal or
sys.exit (it re-execs the process instead).
This meant that we didn't have a mechanism for testing the event queue
dump/load functionality in the development environment. We fix this
by adding such dumping/loading. However, this breaks the automatic
reloading of open browser windows on a server restart, so we add that
back in by adjusting the special `restart` events to pass a special
`immediate` flag when used in development.
This also has the benefit of removing the "Bad event queue" errors one
would get on every file save induced restart on the Python console.
Apparently it isn't always the case that removal of jquery and the DOM
prevents cleanup_event_queue from being called via the postunload
hook, so add a check to avoid it being double-called.
Previously, the browser might restart a get_events operation even
while it was in the middle of executing a `DELETE /events` query to
cause its event queue to be de-allocated. This was a rare race
condition when we weren't notifying clients when event queues were
de-allocated, but this will become a common case in the next commit.
This is a no-op right now, but we'll want the new structure for the
next commit, and splitting this out makes it a lot easier to read what
is actually changed in the next commit.
This should fix a problem we've been having with errors downloading
the PhantomJS packages from their original hosting service.
Eventually we should move it to an S3 bucket.
cd2348e9ae broke installing Zulip in
production since it didn't correctly update the puppet configuration
to call the process_queue script using the new argument format.
This commit isn't ideal in that I'd prefer to not require updating
puppet in sync with the actual running code, but we don't have a great
mechanism for doing that.
Fixes#586.
Previously, we needed to update the installation instructions with the
current version of Zulip in production every time we did a release,
which was kinda a pain (and hadn't happened since 1.3.6).
Fixes#576.
[commit message details expanded by tabbott]
The original logic for incremental presence list updating from
668d0d9dfa incorrectly attempted to
insert the user 1 spot later than its proper index in the listing.
Now that we're doing presence updates in a performant fashion, we
don't need to throttle processing these events, and in fact the
throttling of these events created a correctness problem, since we're
now doing incremental updates rather than just rerendering everything
after each event.
The code in 668d0d9dfa for removing an
existing user from the user list to update the status didn't correctly
quote the email address of the user in its jquery selector.
While we already don't link to /terms anywhere on the site, they can still be
accessed if you navigate to /terms directly. Now, those routes will only be
exported on the Zulip.com service.
We should ideally provide a mechanism for deployments to specify their own
terms without modifying source code; in the interim, sites that have already
customised the provided Zulip.com terms can simply carry a patch reverting this
commit.
This change drops the memory used for Python processes run by Zulip in
development from about 1GB to 300MB on my laptop.
On the front of safety, http://pika.readthedocs.org/en/latest/faq.html
explains "Pika does not have any notion of threading in the code. If
you want to use Pika with threading, make sure you have a Pika
connection per thread, created in that thread. It is not safe to share
one Pika connection across threads.". Since this code only connects
to rabbitmq inside the individual threads, I believe this should be
safe.
Progress towards #32.
Ryan Moore
Tim Abbott
Tim Abbott
Ashish
goelakash
Eklavya Sharma
akashnimare
SummerBulb
Kunal Gupta
Kartik Maji
Vladislav Manchev
Zev Benjamin
Luke Faraone