New dependency: sockjs-tornado
One known limitation is that we don't clean up sessions for
non-websockets transports. This is a bug in Tornado so I'm going to
look at upgrading us to the latest version:
https://github.com/mrjoes/sockjs-tornado/issues/47
(imported from commit 31cdb7596dd5ee094ab006c31757db17dca8899b)
This may require just doing an mv on the home directory, plus changing
the home directory in /etc/passwd. It should of course be done carefully.
(imported from commit 660997d897ee6d33563af74f0fc5d4267a911755)
This requires doing an `mv` and then `puppet apply` on each of staging
and prod as part of the deployment process.
(imported from commit 5d0be64a3846f7151d2036d2e0b31049bc1c2dd2)
You can immediately apply the changes by passing '-f' or '--force' as
the first argument.
(imported from commit 91f00f6afd3c42e2b11f60e92fc20962bc952f0d)
This temporarily breaks the rabbitmq consumer checks for the
user_activity and notify_tornado queues because their state files
were renamed to match their queue names. It will be fixed for
staging in the next commit.
(imported from commit a6aaa330a1134d8ddffe8f4959deb12b219f241a)
The minify logic doesn't have an easy way to detect that you
removed a file since the last deployment.
(imported from commit 50d05fcdad382a586073c06d29d279433d1bba81)
people_list and people_dict include the feedback bot and anyone you've
cross-realm PM'd with. Useful for autocomplete, but not for admin and
stream settings views.
Fixes the UI part of Trac #1772.
(imported from commit cdefd4e86980447aad5190e7fc8ae3666d66e3c3)
This shows up when you're not running a Zephyr mirroring bot and lets
you use Webathena to have us run it. Obviously needs more docs.
Current problems include:
* supervisorctl reload ends up recreating /var/run/supervisor.sock
with the wrong permissions, so it only works once in a row before
you need to chmod that.
* /etc/supervisor/conf.d needs to be humbug-writeable; this is a clear
local root vulnerability
* This uses SSH and thus is kinda slow.
(imported from commit 7029979615ffd50b10f126ce2cf9a85a5eefd7a2)
This change will allow us to test the muting feature on
staging. Any topic named "muted" will automatically be
muted. You can also mute any other topic on the console:
muting.mute_topic('devel', 'ios');
current_msg_list.rerender();
More UI around this experiment will be coming soon, as well
as support for muting entire streams.
The muting module keeps track of which topics are muted, but a
user can expand muted messages, and once that happens, the
messages are marked with the "force_expand" flag that gets
persisted to the back end.
Muted messages are rendered in similar fashion to the summarized
rows, and as part of unifying some of that code, we have
made it so that expanding a summarized section doesn't remove
individual flags related to summaries; instead, the messages
get the force_expand flag set.
(imported from commit acee4190e63813d46850415c41ff8ebfae4a6953)
Have ui.set_presence_list() only touch the presence list.
Before this change, it was calling update_unread_counts(), which
has a bunch of side effects unrelated to the presence list.
(imported from commit 690f754d78874a03fa36f8ff8765d5a63e431d28)
This was broken in two ways:
1. Commit ad59d6f78042ce89, "Make the left sidebar and right sidebar
more consistent", pushed last Monday, changed the markup for the right
sidebar without changing a selector in stream_list.js that was looking
for the old markup.
2. Even then, whenever new user presence information came in, we would
rerender the user list and blow away the unread counts. This commit
patches around that by updating unread counts after rendering the user
list. I'm not sure what broke this or how it was working before.
(imported from commit 53ed40139e257e44411e918d1ecdce3a49e9ee51)
The functions add_dependencies() and set_global() are convenience
methods that allow you to modify the global namespace while
the current file is running but then have it be cleaned up
by index.js when you're done.
(imported from commit f75b8a10c19f773a8d2d3a8fa4bc39b1679566fe)
To use, make sure your Node installation is up to date and install
istanbul with npm:
sudo npm install -g istanbul # 'sudo' may be optional on OSX
Then run
tools/test-js-with-node cover
and navigate to coverage/lcov-report/js/index.html in a browser.
(imported from commit 2da4894d1725e2f9540b3895304246e3cd138f6c)
There are no functional changes; you can still use the shell script
tools/test-js-with-node. It just delegates now to the new index.js to
iterate through all the other .js files in the test directory and run
them. This sets the stage for Istanbul to correctly compute test
coverage.
(imported from commit 6f521c78b7a314d010fa113f9c2c971ab999b637)
Users were getting confused about why the unread count in the sidebar
/ notification bar / Dock was different from what the bankruptcy modal
said, so only show them the true server count until they've made a
decision.
(imported from commit 71d376cd4a85749ccf49936b251e6b8ac21361b7)
Move zulip.subject_dict into composebox_typeahead.seen_topics,
and encapsulate the use of seen_topics inside composebox_typeahead
with add_topic() and topics_seen_for().
(imported from commit 2bc2d1714fabdc07a661cbf815d14b36a08990e2)
We now show a list of users and allow you to deactivate a user using the
same process as `python manage.py deactivate_user`.
We add a new menu item accessible from the gear icon which will eventually
have much more than just this, but we have a good start here.
Here we also add a property to UserProfile which determines whether you're
eligible to access the administration panel, and then have code which shows
the menu option if so.
This introduces a new JS file, admin.js.
(imported from commit 52296fdedb46b4f32d541df43022ffccfb277297)
We instead implemented the ~desired functionality here using the
API and a bot to make a totally read-only, static, slowly-updating
view into the Zuliverse.
This is the moral equivalent of reverting deb035b4c702fcdb0e660ed549fe74c682abb6d9
(imported from commit 9d743fe82f197b37f005e5a038f77cc4b8566024)
1) The class Filter now lives in its own module.
2) The function canonicalized_operators() is now a class method on Filter.
3) The function message_in_home moved to filter.js and became private.
4) Various calling code had to change, of course.
5) Splitting out Filter helped simplify a few tests.
(imported from commit e41d792b46d3d6a30d3bd03db0419f129d0a2a7b)
The compose_fade has three public exports:
set_focused_recipient
unfade_messages
update_faded_messages
All code was pulled directly from compose.js, except for the
one-line setter of set_focused_recipient. The focused_recipients
variable that used to be in compose.js was moved to compose_fade.js,
hence the need for the setter.
(imported from commit 462ca5d0d0bd58612d0197f3734a8c78de8c6d30)
"Kiosk mode" is a "read-only" Zulip suitable for embedding into
an iframe on another site. I say "read-only" in quotation marks,
because the account is still a fully-fledged active account on
the server, and we just tear out a bunch of stuff in Javascript
(that a malicious user could easily re-enable).
So in that sense, it's not actually safe in security-sensitive
environments -- malicious users logged in via kiosk mode
can do anything the kiosk-mode user can do.
(We need this functionality for the customer3 realm specifically;
we'll possibly just tear this code back out once that experiment
has run its course.)
(imported from commit deb035b4c702fcdb0e660ed549fe74c682abb6d9)
util.enforce_arity takes a function and returns a new version which
throws an error if an incorrect number of arguments (as determined by
the function prototype) are passed.
(imported from commit 20e69a6dc7b6f8455726ab4fae8d5b7b04dc4103)
The test-all script now calls the symlink, and the run script
has been cleaned up to be symlink friendly.
(imported from commit 8abb5c1e5744416e94ff843e50c53e0d0f7e1316)
This is a pure refactoring that mostly just moves code from
subs.js to the new stream_color.js and updates module references
accordingly. In order to prevent introducing some exports,
update_stream_color was given an additional "sub" parameter
and update_stream_sidebar_swatch_color was given an "id"
parameter.
Killed off unused initial_color_fetch var.
(imported from commit b7644ce67f50d31fb46f564d758d661eea776aa6)
This includes a hack to preserve humbug/backends.py as a symlink, so
that we don't need to regenerate all our old sessions.
(imported from commit b7918988b31c71ec01bbdc270db7017d4069221d)
This needs to be deployed to both staging and prod at the same
off-peak time (and the schema migration run).
At the time it is deployed, we need to make a few changes directly in
the database:
(1) UPDATE django_content_type set app_label='zerver' where app_label='zephyr';
(2) UPDATE south_migrationhistory set app_name='zerver' where app_name='zephyr';
(imported from commit eb3fd719571740189514ef0b884738cb30df1320)
For now, we just get emails about referrals that we have to follow up
on manually.
I don't love using the name "referral" in some places and "invite" in
others, but we already use the verb "invite" to mean something else
and "invite" is a canonical noun.
(imported from commit 0814c18395952fcdef234c1584984f71ca1b6f37)
We always intended to move the pointer up when you were at the
top of the viewport and scrolling up, and we always intended to
move the pointer down when you were at the bottom of the viewport
and scrolling down. We didn't intend to move the pointer up
when you were at the bottom, or vice versa. This commit fixes
that bug, simplifies the code, and inlines the code into ui.js.
(imported from commit 77e2ace9d2fc1025e1349e3be13c76c3a397fd38)
This follows up on extracting code to js/search_suggestion.js, and all
the search tests apply to the new module.
(imported from commit dec6c8614c25b4f82da57edeaddc7cfef28260a5)
Moved 400+ lines out of search.js into search_suggestion.js. This
leaves search.js primarily responsible for lots of little DOM
interactions, whereas search_suggestion is more about data.
(imported from commit 53d08b29367c0172e483064f213538d45098279e)
I added our "static" directory to NODE_PATH for our JS unit tests.
This eliminates most of the verbosity in our require statements, but
it still requires us to explicitly call out "js" or "third"
subdirectories, which should make it a bit easier for folks reading
the tests to distinguish modules from js, third, or node itself.
(imported from commit b77a5283135d388d46f4b7e511acc59986f1a8ba)
In the future, this could come from the server to enable individual
experiments on a per-realm basis.
(imported from commit 8fd1ba1910b3cfd131f58bab8efbd11a42053bc3)
Previously, when the first `which` failed, the entire script would
fail with exist status 1 because we pass `-e` to the interpreter.
(imported from commit 601de3b3e3edd90110fc478f7874e644009d1b62)
This covers most of the module's functionality, with special
emphasis on lines that use underscore.js (_.each and _.filter).
(imported from commit 074181a0273286a258504be634bdd1cead2eecd5)
I haven't hooked this into test-all yet, but I did modify
message_tour.js to accomodate the test.
(imported from commit e58f595f4fe1160f539c18ec09dbe22eebf1f104)
When you return to Home, we normally restore your last
position when in Home, which might have been before the
first unread message. This is cumbersome for sidebar readers,
so now we keep track of messages that you visit while on a
tour away from Home, and when you return Home, we skip forward
any messages that were in the tour, landing on the last visited
message. This all happens before rendering.
(imported from commit 9124a231d94f153e283e5ea95e40c50a58406275)
When we push this to staging, we'll need to rename the bot in the
database and also pull on git.zulip.net.
(imported from commit 22b2397b197c8820f0e55daecd8f98d829e195bd)
Created up, down, page_up, page_down, to_home, and to_home pretty
mechanically from the old hotkey.code.
(imported from commit 5956b91c2e0122c6440f70db9b92f918c9b599aa)
* CC to code-review@zulip.com
* If domains are omitted, assume @zulip.com
* Don't set email or site explicitly, instead relying on ~/.humbugrc
(imported from commit 9700a0e3d6aac97c8030dcbaba7790018173929a)
We only want to change cases where we're talking about the hostname; HTTP
requests should still go to staging.humbughq.com for now.
Before this commit is deployed the hostname of staging.humbughq.com should
be changed to staging.zulip.net on the VM.
(the same for prod)
(imported from commit 7412530773f720ac227f40061c9ddb1a851e19bb)
This is a big change affecting lots of areas:
* Pipeline no longer deals with JS (though it still minifies CSS)
* A new script, tools/minify-js (called from update-prod-static),
minifies JavaScripts
* A command-line argument --prev-deploy, if passed to minify-js or
update-prod-static, is used to copy minified JS from a previous
deploy (i.e., a previous git checkout), if the source files have
not changed
* update-deployment passes --prev-deploy
* Scripts are now included with the minified_js template tag, rather
than Pipeline's compressed_js
Also, as a side benefit of this commit, our Handlebars templates will
no longer be copied into prod-static/ and accessible in production.
Unminification is probably broken, but, per Zev and Trac ticket #1377,
it wasn't working perfectly before this change either.
(Based on code review, this commit has been revised to:
* Warn if git returns an error in minify-js
* Add missing output redirects in update-prod-static
* Use DEPLOY_ROOT instead of manually constructing that directory
* Use old style formatting)
(imported from commit e67722ea252756db8519d5c0bd6a421d59374185)
The goal here is to make it easier to do ad hoc profiling on
our codebase, particularly by running tests.
(imported from commit 71da06feb3a369dec8dc4d8391f7f40e4c2d02ff)
Caveats:
- Since Chrome has trouble using W3C Notification when it's not
initiated by a user gesture, we try to use webkitNotification first.
- FF doesn't allow iconUrl to be of a different origin, so it won't display
our gravatars
(imported from commit c4f99ce6927a0d203d9f220d50b06737779bd7f8)
Per http://docs.python.org/2/library/os.html#os.getlogin, getlogin()
only works when you have an associated controlling tty.
This script didn't work previously because when we do deployments there
is no tty. Thus, we switch to the alternative mechanism for determining
the current username described on the page linked above.
(imported from commit 1dbcf98fd7248d20e501fd7fb22e1dbd306040fd)
If you're running this as a user other than "humbug" on a deployed server,
you're going to have a bad time.
Specifically, memcached won't work, and other undefined behaviour may
occur.
So here we add a check and error out if you run this script on an
app_frontend as non-"humbug".
(imported from commit a3d5f0f58ded42393c03f4d21b4650494fae418f)
Now that we have more than a month worth of data, it's nice to
actually just display it all.
(imported from commit f2ec027af81699d6937e52b18ac40973a54c349b)
Otherwise we end up running Tornado against the _previous_ version of
our code!
Testing of using supervisorctl to stop and then start a process shows
it takes about the same amount of time as doing a supervisorctl
restart, so there's no reason not to split the two commands apart and
make it super clear that nothing is running at the time that we move
the deployment symlink.
(imported from commit c38049da2bfc9fa94320a32dbf3240d1fcba67f7)
1) When you send a message, restore the focus to the composebox, targeted at the same recipient
2) If the composebox is completely empty and you press up or down, have that close the composebox and take the appropriate action
3) If you started the compose via a reply option (r, enter, click), don't refocus the composebox if the cursor has changed.
(imported from commit 84545e49d06959eb62e7fd2b22e1387383df6d1d)
Does a test-all using /tmp.
Not safe to run with other tests simultaneously.
You use it just like tools/review, and on a successful test-all
it will submit the code review, without tying up your git working directory.
It depends on new patches to the acrefoot-bulk_create_with_id-1.5.1 branch of
django. Updating this is the [manual] step required. Since test-all doesn't happen
on staging or prod, this is not required for this commit to be deployed, but other
django patches will help with the SSL connection errors we've been seeing.
(imported from commit 3fd3ff00240a2d648c4c54748a8a70616067ef7d)
Since update-deployment is run on the host being deployed to and only
has access to a recent clone of the git repository, it doesn't
necessarily have the old refs available for reverts.
(imported from commit 3652f58a7b165c805822bf6d8a4f0792c629e28e)
Previously our receive API bindings were broken in our API tarballs
because we weren't including the receive API bindings which they used.
This requires our deploying the built API tarball to the prod server
when we deploy it so that the link on /api isn't broken.
(imported from commit 14ecaab34556f4e29c72f4f567d8af73c89d6297)
With our currently rather slow deploys, this helps with the confusing
case where 2 deploys happen in quick succession and the second is
blocked by the lock.
(imported from commit 5a8bd449ecbe0825838faf72099b67f94db7797a)
Exit with a failing status if any of your
non-excluded JS/Python files have trailing whitespace.
(imported from commit 0aa184fa85a4a1ddeef6ea40dff57ce43c10e459)
This is a little hackish in that I'm using humbug-send rather than
importing humbug_git_config.py to get the API keys and all, but this
was quick to do and I think it's valuable for us to have this output
on our deploys.
(imported from commit 24d7954d04aadeff77ea9cdb76fdc8be055f3ac5)
Re-focuses on the compose box after a send, but only if
the compose-box was opened by responding to the message
at the cursor (by hitting "r", enter, or clicking on the message)
(imported from commit 8e7560c8ea31397b57b2bc3e2e7d9dd996226a6f)
Modified files need to be copied into the right place. The checkout
on git.humbughq.com also needs to be updated.
(imported from commit dbe9e05a0512e1f59c7819dd8d44c2c4e9c83bcf)
We let supervisor create the socket for us by making humbug-django a
fcig-program. Unfortunately, supevisor doesn't support putting
fcgi-programs in groups (see
https://github.com/Supervisor/supervisor/issues/148), so we have to
restart tornado and django separately.
To deploy, copy the config files over and restart nginx and
supervisor (via stopping and then starting it because restart is
broken). I believe the automated restart as part of
update-deployment will fail because of the way supervisor treats
programs in groups. If so, after restarting supervisor, you will
also need to run restart-server manually to fill the caches and then
delete the lock directory in humbug-deployments.
(imported from commit bfb5db7dd42dcbc4bfefa2944355b3cbb2ef9104)
The amount of process downtime during a supervisord-mediated restart
appears to be linear in the number of processes that are being
restarted. Therefore, restarting just Django and Tornado causes less
downtime than doing them at the same time as the other worker
processes.
(imported from commit 1fa9ef547bcd88caeec49800664e37d5f2fcb7a8)
Currently our test database is backed by sqlite; this commit moves
us to using postgres for our all database needs. This, in conjunction
with the patched django on github, allow us to have fewer hacks and
more true-to-life tests. It also sets the stage for testing the bulk_create
and schema search_path patches made to django.
Developers will need to run:
./tools/postgres-init-test-db
./tools/do-destroy-rebuild-test-database
this is assuming that they have already run:
./tools/postgres-init-db
./tools/do-destroy-rebuild-database
at some point on this pg_cluster. (The ordering is important; it will other-
wise complain about the south_migration table).
(imported from commit c56c6f27e13df7ae10b2e643e65d669dde61af3d)
This change will make it so that processes related to the app.humbughq.com
server are run under supervisord, which uses a state machine model to ensure
that programs are running. It also ensure process startup order.
We will need to manually switch the old way of running server (in screen) into
this new way of doing things, on both staging and prod (app_frontend.pp has been
updated appropriately). This means:
1) cp servers/puppet/modules/humbug/files/supervisord/conf.d/humbug.conf /etc/supervisord/conf.d
2) installing the supervisor package.
3) killing those while loops in that screen session
4) mkdir /var/log/humbug (as root)
5) /etc/init.d/supervisord start
6) check that nothing broke
(imported from commit 055269a70973db89acd69049e01b185fabdc8f90)
Tim Abbott
Zev Benjamin
Steve Howell
Jessica McKellar
Luke Faraone
Waseem Daher
Leo Franchi
Kevin Mehall
Scott Feeney
Steve Howell
Tim Abbott
Kevin Mehall
Zev Benjamin
Luke Faraone
Scott Feeney
acrefoot
Leo Franchi