This helper was added in eac6463031 and
used by the "message.handlebars" file. This is no current call for
this helper in the codebase, hence it is removed to improve coverage.
This commit also marks template.js to have 100% test coverage.
In the openapi specs, the update_message_flags event is documented as
having a `operation` (deprecated) field, alongside the modern `op`.
This causes check_schemas warnings like this:
NEED SCHEMA to match OpenAPI update_message_flags_add_add_event
NEED SCHEMA to match OpenAPI update_message_flags_remove_remove_event
as check_schemas uses both `op` and `operation` for constructing the
event name.
Being deprecated (and really only still there for
backwards-compatibility with the original error of having it present),
`operation` will be removed eventually, therefore we can safely
ignore it from being used in openapi schema validation.
Part of #17568.
This also fixes the suggestions for the following words: disabled,
disables, disabling, implemented, implementing, implements, kept,
made, took, using.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
We basically move all the tests from backend and frontend test
files to zulip-ci workflow. This results in GitHub Actions
nicely displaying all the tests separately.
We have disabled CircleCI and are using GitHub Actions for automated
testing.
docs: Changed context from CircleCI to Github Actions and wrote
some documentation specific to GH Actions.
tools: Replaced env checks for CIRCLECI with GITHUB_ACTION.
README: Use GitHub Actions build status badge.
GitHub Actions supports doing more than just CI,
and so in some contexts it's less obvious that we're
talking about just the CI if we refer to it instead of CircleCI.
This commit sets the system default locale to `en_US.UTF-8` before
running the tests.
This was necessary because when running tests with Firefox, it was
failing on some test verifies the results of locale-aware sorting.
When exception is raised inside an exception handler, Python 3
helpfully prints both tracebacks separated by “During handling of the
above exception, another exception occurred:”. But when we’re using
an exception handler to retry the same operation, multiple tracebacks
are just noise. Suppress the earlier one using PEP 409 syntax.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
./tools/test-all had a short option -f which was used for running it
by skipping provision check. This short option paired well with its
long version --force, but in an effort to provide more clarity over
the use of force option it is renamed to --skip-provision-check.
Following this rename -f option for test-all does not pair well with
its long part which is now --skip-provision-check. Also, this option
is rarely used. So the short form -f is removed in this commit. We
now provide only its longer version --skip-provision-check.
This is a part of commits done to address issue #17455.
This commit changes help message for skip-provision-check argument
used with various test commands. The message is changed with an
intention to be more clear about what this option actually does.
This commit is in series of various changes done to provide clarity
over the use of --force option which is renamed to
--skip-provision-check.
Fixes: #17455
This commit renames --force argument used with various tests to
--skip-provision-check. As a consequence of this name change all other
files that set --force option for the test commands have been updated.
This change is done in order to provide more clarity for using this
option for runnning tests.
This commit addresses issue #17455.
This commit moves --force option used with various tests to
test-scripts.py to have it alongside the logic that does provisioning
status assertion.
This is a step towards providing more clarity over use of this
argument with tests as asked in issue #17455.
The "Narrow to PM with" notification above the composebox was
double-escaped, mangling names with single quotes in them. This removes
the escaping in i18next, causing the name to be escaped only in
handlebars.
Replaced methods/functions of moment.js with date-fns library.
The motive was to replace it with a smaller frontend timezone library.
Date-fns ~ 11.51 kb
moment.js ~ 217.87 kb
Some of the format strings change because date-fns encodes them
differently from how moment did.
Fixes#16373.
We often send only one field (away or status_text)
to be updated.
So we have to make our schema support optional
keys.
As a result of the more flexible schema, we no
longer need to exempt the node fixtures from
our schema checks.
Allowing any admins to create arbitrary users is not ideal because it
can lead to abuse issues. We should require something stronger that
requires the server operator's approval and thus we add a new
can_create_users permission.
f82cc4ed06 started checking all
zulip/zulip GitHub links in CI. Instead, it should have checked only
zulip/zulip file and directory links since checking other
links require making requests to GitHub servers.
This fixes a bug where the typeahead did not include the
zulip specific langs (such a `quote`, `spoiler` and `math`)
as these weren't passed to the typeahead's source.
Introduced in af64c52166.
Fixes#16862.
I added these hooks in Zulip Desktop 5.5.0; handling these events in
the frontend will let us remove the janky desktop-side fallback code
that uses fake click events on menu items with specific indexes.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Steve asked me to remove this, since the tictactoe game was always
intended as a proof of concept. Now that we have poll and todo
widgets, the sample code for tictactoe has much less value.
We replace the content and type in test_widgets.py to maintain
coverage.
There is only one PostgreSQL database; the "appdb" is irrelevant.
Also use "postgresql," as it is the name of the software, whereas
"postgres" the name of the binary and colloquial name. This is minor
cleanup, but enabled by the other renames in the previous commit.
This moves the puppet configuration closer to the "roles and profiles
method"[1] which is suggested for organizing puppet classes. Notably,
here it makes clear which classes are meant to be able to stand alone
as deployments.
Shims are left behind at the previous names, for compatibility with
existing `zulip.conf` files when upgrading.
[1] https://puppet.com/docs/pe/2019.8/the_roles_and_profiles_method
There was likely more dependency complexity prior to 97766102df, but
there is now no reason to require that consumers explicitly include
zulip::apt_repository.
Installing an updated linux kernel package, as can happen during the
`apt dist-upgrade` done by the installer, can cause grub to pop up a
prompt to update its configuration file. In an unattended headless
configuration, this will stop the installation.
Explicitly configure apt to be non-interactive, and prefer the newest
configuration, during the install.
We used to send occupy/vacate events when
either the first person entered a stream
or the last person exited.
It appears that our two main apps have never
looked at these events. Instead, it's
generally the case that clients handle
events related to stream creation/deactivation
and subscribe/unsubscribe.
Note that we removed the apply_events code
related to these events. This doesn't affect
the webapp, because the webapp doesn't care
about the "streams" field in do_events_register.
There is a theoretical situation where a
third party client could be the victim of
a race where the "streams" data includes
a stream where the last subscriber has left.
I suspect in most of those situations it
will be harmless, or possibly even helpful
to the extent that they'll learn about
streams that are in a "quasi" state where
they're activated but not occupied.
We could try to patch apply_event to
detect when subscriptions get added
or removed. Or we could just make the
"streams" piece of do_events_register
not care about occupy/vacate semantics.
I favor the latter, since it might
actually be what users what, and it will
also simplify the code and improve
performance.
I think it's important that the callers understand
that bulk_add_subscriptions assumes all streams
are being created within a single realm, so I make
it an explicit parameter.
This may be overkill--I would also be happy if we
just included the assertions from this commit.
In addition to being generally more correct, this works around a bug
in Node.js that causes webpack-dev-server to corrupt the terminal
state when exiting as a background process.
https://github.com/nodejs/node/issues/35536
Signed-off-by: Anders Kaseorg <anders@zulip.com>
It happens that whether you add a reaction or remove
a reaction, we send the exact same fields, just using
a different op code.
This sort of symmetry is actually kind of rare, as
usually "add" events have more fields, and "remove" events
might just send an id of something to remove.
Our openapi schema treats these as two seperate events,
so we are more consistent with it, and it helps our
schema-checking tooling for node fixtures, too.
Note that we now have to exempt the two events from
our openapi checks, due to the is_mirror_dummy field
in the deprecated user block. We can decide how to
handle this later--one possibility is to just add it
as an optional field on the event_schema side.
Note that we make the schema for profile_data
slightly more realistic, but it doesn't actually get
exercised by our current tests (apart from
making sure it's a dict), since we don't have
profile data for our test realm.
We also don't have the optional fields for bots,
since our tests don't exercise that, nor
delivery_email.
So we exempt realm_user_add_event from openapi
checks for now.
When we try to match the openapi specs better, we
will probably want to add a few tests to test_events.
Obviously getting good coverage for adding users
would be nice for all these scenarios:
* delivery_email matters
* bots
* realm has profile fields
This is a prep commit for supporting "presence"
events, where the key of the dictionary is some
arbitrary string like "website" but the value
of the dictionary is another dictionary itself
with keys that are more like variable names.
This also forces us to create TupleType.
We exempt this from the openapi check,
since we haven't figured out how to model
tuples in openapi with the same precision
as event_schema (and it may be impossible).
Long term we just want to stop dealing in
tuples, of course.
StringDict is a data type for representing dictionaries where
all keys and values are strings. Add this data type to data_types.py
and edit other files so that this data type is put to use and tested.
(slightly tweaked by @showell to remove a comment and shorten
a var name now that we have a proper data type)
We also make our schema in event_schema reflect this,
which in turn makes us match the already accurate
openapi spec, so we no longer need to exempt four
types of events from our sanity checks.
We might want to rename the tool to something more
general now, since we are really reconciling three
things:
- node fixtures
- event_schema checkers for test_events
- openapi specs
The way we compare python and openapi schemas is
as follows:
- first convert openapi schemas to be build
from DictType, ListType, etc. with from_opeapi
- do a diff on the schemas
Most of the new code is just having the FooType
family of classes serialize themselves with schema().
Defining types with an object hierarchy
of type classes will allow us to build
functionality that was impossible (or
really janky) with the validators.py
approach of composing functions.
Most of the changes to event_schema.py
were automated search/replaces.
This patch doesn't really yet take
advantage of the new FooType classes,
but we will use it soon to audit our
openapi specs.
SimpleBar 6.0.0-beta.2 through -beta.6 are built with ES6 syntax (I
assume inadvertently: https://github.com/Grsmto/simplebar/issues/523),
and its latest tag has moved back to 5.2.1 anyway.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
user_profile will be None for web_public_guests here. Hence, for
settings (of which most be inaccessible by web public guest),
which require a user_profile, we either set an empty value for
them or set them to a default value. This will help render
the frontend or extend support to our clients without breaking
a lot of code.
Tweaked by tabbott to add many comments.
The reason higher expected_time_to_clear_backlog were allowed for queues
during "bursts" was, in simpler terms, because those queues to which
this happens, intrinsically have a higher acceptable "time until cleared"
for new events. E.g. digests_email, where it's completely fine to take a
long time to send them out after putting in the queue. And that's
already configurable without a normal/burst distinction.
Thanks to this we can remove a bunch of overly complicated, and
ultimately useless, logic.
The race condition is described in the comment block removed by this
commit. This leaves room for another, remaining race condition
that should be virtually impossible, but nevertheless it seems
worthwhile to have it documented in the code, so we put a new comment
describing it.
As a final note, this is not a new race condition,
it was hypothetically possible with the old code as well.
We need this information in the frontend to:
* Display the 'view in playground' option for locally echoed messages.
* When we add a UI settings for realm admins to configure their
playground choices, we'll need to use these canonicalized aliases
for displaying the option.
Hence, this tweaks the tool which generates pygments_data.json to contain
the data we need.
Bumping major PROVISION_VERSION since folks need to provision in both
directions.
Tests amended.
css-loader@4 broke @import statements referencing files with
extensions other than .css, unless those @import statements are
compiled away by another loader. Upstream is more interested in
arguing that such @import statements are semantically incorrect than
applying the one line fix.
https://github.com/webpack-contrib/css-loader/issues/1164
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This was a broken abstraction that returned to its caller within
multiple forked processes on exceptions, and encouraged ignoring the
error code (as all of its callers did).
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
The specification says “any sibling elements of a $ref are ignored”,
so their presence, although not invalid, indicates a mistake. yamole
incorrectly merges these siblings into the referenced object, but we
should not rely on this nonstandard behavior.
https://swagger.io/docs/specification/using-ref/#sibling
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Any exception is an "unexpected event", which means talking about
having an "unexpected event logger" or "unexpected event exception" is
confusing. As the error message in `exceptions.py` already explains,
this is about an _unsupported_ event type.
This also switches the path that these exceptions are written to,
accordingly.
8e10ab282a moved UnexpectedWebhookEventType into
`zerver.lib.exceptions`, but left the import into
`zserver.lib.webhooks.common` so that webhooks could continue to
import the exception from there.
This clutters things and adds complexity; there is no compelling
reason that the exception's source of truth should not move alongside
all other exceptions.
There are three functional side effects:
• Correct an insignificant but mathematically offensive bias toward
repeated characters in generate_api_key introduced in commit
47b4283c4b4c70ecde4d3c8de871c90ee2506d87; its entropy is increased
from 190.52864 bits to 190.53428 bits.
• Use the base32 alphabet in confirmation.models.generate_key; its
entropy is reduced from 124.07820 bits to the documented 120 bits, but
now it uses 1 syscall instead of 24.
• Use the base32 alphabet in get_bigbluebutton_url; its entropy is
reduced from 51.69925 bits to 50 bits, but now it uses 1 syscall
instead of 10.
(The base32 alphabet is A-Z 2-7. We could probably replace all of
these with plain secrets.token_urlsafe, since I expect most callers
can handle the full urlsafe_b64 alphabet A-Z a-z 0-9 - _ without
problems.)
Signed-off-by: Anders Kaseorg <anders@zulip.com>
It was broken by commit aaedec1fdb which
moved it to tools/i18n without adjusting its relative path references,
and it contains a sketchy injectable os.system call that I’d like to
remove.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
In 6653e19e3a we added
the convenient line to tell folks about the coverage
report. But if we failed coverage checks, we didn't
show the link.
Arguably we should just always show this, even if
tests fail, but that can also be potentially confusing.
The code to run single files was added
in c15695e514,
and it's just kinda strange code.
We already do a lot of file logic in Python
to check for line-coverage, so it's easier
to just have all the logic in Python.
This adds a new feature--you can now specify
the actual file:
./tools/test-js-with-node frontend_tests/node_tests/people.js
(This is helpful if you just want to use
shell autocomplete.)
Another minor change is that if you specify
individual files, we won't sort them. This is
important when you're trying to hunt down test
leaks.
Finally, we have a nicer message if we can't find
the file.
nyc was added in 29f04511c0
All the stuff after "&&" was actually passed to
node, because we didn't use shell=True, so the
"nyc report" command didn't run, and the ugly
finder.js code just skipped over all the final tokens.
We lost the war against top level configuration files many moons ago.
This is what developers and tools expect. And it seems to be required
for eslint-import-resolver-webpack (there’s ostensibly a {"config":
"tools/webpack.config.ts"} option, but it doesn’t work correctly:
https://github.com/benmosher/eslint-plugin-import/issues/1861).
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This fixes the regression introduced in the pervious
commit to regain the 100% line coverage in `user_pill.js`
as well as `stream_pill.js`.
The new `stream_edit.js` mainly tests for:
* The stream related queries of the typeahead in `user_pill.js`
* The "Add subscribers" event handlers.
* The event handler which displays the settings for a stream.
We update the pills typeahead logic to also include
stream results and pass the "stream" key in `opts`
to enable this option for the Add subscriber form.
This commit implements the feature of adding all the
subscribers of another stream in the "Add subscribers"
UI, with the help of a new "stream_pill.js` file.
We temporarily add `user_pill.js` to the EXEMPT_FILES
list as typeahead will be set up in `stream_edit.js`
file which does not have any dedicated tests file.
Work towards #15186.
We merge bootstrap-responsive.css into bootsrap.css since that is
how bootstrap distributes it from this version onwards.
bootstrap.js has a lot of changes to it which completely breaks
our typeaheads and popovers, so we will have to override these
plugins with our version of these plugins. In future versions
of bootstrap when we use npm, we can just choose not to
import them.
The node package allow use to control xvfb apt package in puppeteer
tests. This help us create a fake display so we can run puppeteer in
headful (headless: false) mode, which is required to use the chrome
extension desktop capture API.
Now that all casper tests have been migrated to
puppeteer, there's no need for having casper
related things.
Removed the casperjs package and removed/replaced
casper in few places with puppeteer.
Only removed few of them which I'm confident
about. Also didn't make any changes in docs
as it would be easier to remove them while
adding puppeteer docs.
This rule is a bit marginal, in that we've only seen this mistake
once, but it is really subtle and took a while for translators to
notice it, so seems worth linting for anyway.
This commit adds automatic detection of extra output (other than
printed by testing library or tools) in stderr and stdout by code under
test test-backend when it is run with flag --ban-console-output.
It also prints the test that produced the extra console output.
Fixes: #1587.
There are file sharing issues with the macOS 10.15.6 and
vagrant. var/remote_cache_prefix was an empty file when using
VirtualBox and Docker on macOS.
Using parallels as a provider for vagrant fixes the issue.
Use --watch-poll which makes webpack to recompile
automatically on file changes, since inotify is not
working here too.
These weren’t wrong since orjson.JSONDecodeError subclasses
json.JSONDecodeError which subclasses ValueError, but the more
specific ones express the intention more clearly.
(ujson raised ValueError directly, as did json in Python 2.)
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Since our Webpack config passes pre-minified JS files to
script-loader, they can’t be used as modules. Use the normal
unminified version, letting Webpack minify it and give us source maps.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
The hash keys were missing hash for package.json and yarn.lock
because they were not present since we don't do a full checkout
in this job. We fix this by sending over those files and generating
hashes from them.
I usally verify these cache keys by clicking the Restore <cache>
step dropdown menu and then clicking the Run ... dropdown menu again
to see the generated hash.
ES and TypeScript modules are strict by default and don’t need this
directive. ESLint will remind us to add it to new CommonJS files and
remove it from ES and TypeScript modules.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Prior to commit eb4a2b9d4e the center
area of the navbar was based on a structure that appended crumbs or
"tabs" as <li>s, forming a tab_bar and a tab_list.
However, in eb4a2b9d4e we apply a new
style and structure to the navbar which lets go of the convention of
tabs. Hence, we'd like to purge the tab_bar and tab_list labels from
our code base.
We purged tab_list in 1267caf5009118875f47fdafe312880af08024e1.
This commit purges tab_bar, it includes:
- A blanket search and replace of tab_bar with message_view_header.
- Splitting a single line comment in
tab_bar.js / message_view_header.js.
- The renaming of tab_bar.js to message_view_header.js.
- The renaming of tab_bar.hbs to message_view_header.hbs.
- A blanket search and replace of tab_data with
message_view_header_data.
- Replacing the single occurrence of tabbar with message_view_header
(it was within a comment.)
There were a lots of flakes in CI recently because typeahead didn't
appear when Enter was pressed and real emails are not accepted as
valid inputs. To fix this we wait for typeahead to appear and then
click that instead of Enter. We also use delay option to type the
email (100ms delay between keypresses) since without we'd also get
flakes.
Re-enable puppeteer test in CI after this fix too.
😛 should be the most general version, which is the one
with open eyes. Other apps do the same and it also means that :P, which
is converted to 😛 is rendered like the emoticon.
Fixes#15970.
This is used rarely enough that it’s easier to document how to use it
as a non-global than to document the horrifying things that might go
wrong as a global.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This makes sure out fixture data for node tests
is realistic, according to the schemas in
zerver/lib/event_schema.py.
Note that we are still in the process of extracting
schemas from test_events.py -> event_schema.py,
so the checks here are somewhat incomplete as of
now.
One nice thing is that the program will tell us
what checkers are missing, so this can motivate
us to move more checkers to event_schema.py.
I considered just making this happen as part of
tools/test-js-with-node, but it's convenient to
run by itself. Also, it currently requires
Django (although we could fix that), which makes
it just expensive enough that I wouldn't want
to always run it before the node tests.
The previous steps for standing up a new host were somewhat manual.
This further scripts the process, by using the AWS CLI to start the
instance, and pass it a "user data" script to provision itself upon
boot. This results in a hands-off provisioning process which
completes in 5min.
Additional settings are required for `~/.zulip-install-server.conf`.
It is not suited for all roles, as it assumes one instance type and
security group value. Additionally, not all of the post-provision
process is currently automated -- Nagios SSH key verification, for
instance, is still a manual step. There are also additional steps for
database or frontend servers. Regardless, this is a move toward
automated provisioning.
Including anon=1 in API requests will retrieve all contributors
of the repo. If there is no asscoiated GitHub account present for
the commits then the email and name of the author mentioned in
commit messages is returned.
Previously, we copied them to /tmp and from there we specified those
assets we copied in circleci config in presist_to_workspace step.
Copying it to a directory allows us to get rid of list in circleci
config and GitHub Actions's upload artifact (their version of
presist to workspace) doesn't allow us to specify indivivual files
so only is this cleaner but required.
As of commit 87e72ac8e2 (#15267), we
need to be an owner for some of the tested functionality, not just an
administrator.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Zulip converts :) to the 1F642 Unicode emoji and promotes the same emoji
in the popular section of the emoji picker.
Previously Zulip has labeled 1F642 as "slight smile". While that name
conforms to the Unicode standard (which describes the code point as
SLIGHTLY SMILING FACE), it didn't match our use case of the emoji.
If a user types :) or selects the first smile in the emoji picker they
probably mean to express a regular "smile" and not a "slight smile",
which raises the question why they are only smiling slightly.
This commit relabels 1F642 as 😄 and our previous 😄 263A as
:smiling_face:. Note that 263A looks different in our three supported
emoji sets, so it is not suited to be our "default smile".
This change does not require a migration since our emoji system stores
both unicode points and names and handles name changes transparently.
Prettier would do this anyway, but it’s separated out for a more
reviewable diff. Generated by ESLint.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Prettier would do this anyway, but it’s separated out for a more
reviewable diff. Generated by ESLint.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
A few major themes here:
- We remove short_name from UserProfile
and add the appropriate migration.
- We remove short_name from various
cache-related lists of fields.
- We allow import tools to continue to
write short_name to their export files,
and then we simply ignore the field
at import time.
- We change functions like do_create_user,
create_user_profile, etc.
- We keep short_name in the /json/bots
API. (It actually gets turned into
an email.)
- We don't modify our LDAP code much
here.
Fixes#12868.
We now also include python version in the format
'major.minor.patchlevel', when generating hash for a
requirement file. This was necessary since packages tend to
break on different versions of python, so it is important to
track the version on which the venv was setup.
WARN: This commit will force all zulip venvs to be recreated.
success-http-headers-bionic.txt and success-http-headers-focal.txt
differ only in the nginx version so this substitution will allow
us to have single file for both of them. Also this change helps
to avoid CI failure if Nginx version is updated in the OS.
The installer does not adjust the node name if the rabbitmq already
exists, and the default node name bakes in the
`zulip-install-bionic-base` hostname. As such, the resulting LXC
image does not properly start rabbitmq.
Remove rabbitmq, allowing the installer to install and configure it
with a nodename of `zulip@localhost`. This also lets the installed
image be successfully copied and booted under a new hostname without
breaking rabbitmq.
Doing service memcached start instead of restart fixed an issue on
focal build in GitHub actions, where it exits with code 1 when it
is done twice.It is done first in Install Dependencies step and then
again in last step where we call tools/ci/setup-backend again which
runs provision.
Furthermore, I don't belive there is a technical reason we use
restart over start; rather I think it was just a random choice with
the intend to just start the services in CI. I traced the code back
to commit 1f2f497cab if it helps.
Looking at the source code of memcached, the step that's failing is:
start-stop-daemon --stop --quiet --oknodo --pidfile $PIDFILE
which is equivilent to: service memcached stop, we can rule out the
service memcache start since it works. Ideally, we do figure out and
solve the issue of why memcached fails when executing service
memcached stop but I am not equipped with debugging it. And this
workaround seems reasonable rather than a "hacky" solution.
For the relevant code in memcached see:
https://github.com/memcached/memcached/blob/master/scripts/memcached-init.
Finally, the change to the rest of services is for consistency.
According to @showell:
> All the slow decorators can die. That was a failed experiment of
> mine from 2014 days. I have meaning to kill them for a couple years
> now. I wrote this with the best of intentions, but I believe it's
> now just cruft. We never made a "fast" mode, for one. And we kept
> writing more and more slow tests, haha.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Rename rest of function names, classes and comments containing bugdoown
to markdown in test_markdown.py. Also change the refactored classes and
functions occurences in other files.
This commit is part of series of commits aimed at renaming bugdown to
markdown.
Rename the file and all the refrences to file and module test_bugdown.py
to test_markdown.py.
This commit is part of series of commit that renames bugdown to markdown.
This commit is first of few commita which aim to change all the
bugdown references to markdown. This commits rename the files,
file path mentions and change the imports.
Variables and other references to bugdown will be renamed in susequent
commits.
Ubuntu 20.04 "focal" comes up to runlevel 5 several seconds before it
is able to successfully resolve hosts, causing `prepare-base` to fail
while fetching from the apt repositories.
Add an additional check to verify that outbound networking is running
before returning from `lxc-wait`.
As in the previous commit, we can no longer pre-install the wrong
version of postgres. Unfortunately, this leaves it out of the base
image and thus makes testing installs longer.
A generator that yields values without receiving or returning them is
an Iterator. Although every Iterator happens to be iterable, Iterable
is a confusing annotation for generators because a generator is only
iterable once.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Add arrow key navigation support for recent topics.
Simple jquery is used to allow navigation for filter buttons,
a grid system is used for navigation inside table.
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, --interactive used to run tests from the start on a
repeated run triggered when tests failed and we decide to trigger
a re-run by pressing Enter key. Rerunning passed tests is of no interest.
It also used to run all tests in a loop even if all pass.
This commit fixes those both issues i.e it runs again from the
test that failed on pressing Enter and exits if all tests pass.
This fixes a bundle of issues where we were missing "" around
attributes coming from variables. In most cases, the variables were
integers or fixed constants from the Zulip codebase (E.g. the name of
an installed integration), but in at least one case it was
user-provided data that could potentially have security impact.
This adds support for a "spoiler" syntax in Zulip's markdown, which
can be used to hide content that one doesn't want to be immediately
visible without a click.
We use our own spoiler block syntax inspired by Zulip's existing quote
and math block markdown extensions, rather than requiring a token on
every line, as is present in some other markdown spoiler
implementations.
Fixes#5802.
Co-authored-by: Dylan Nugent <dylnuge@gmail.com>
The automated tests running in CircleCI don't actually use the `zulip`
db, so we can skip running migrations on it in some CircleCI shards to
save time.
NOTE: This only effects build jobs that run provision, except the
`production-build` job where we skip building the dbs altogether.
Migrations still run on `focal-backend` build job to ensure
we are testing all our development setup code.
These files can’t use f-strings yet because they need to run in Python
2 or Python 3.5.
Generated by pyupgrade.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
The psycopg2.SQL API unfortunately doesn’t work with
django.db.migrations.RunSQL, so we need to take a detour into
PL/pgSQL for EXECUTE and format.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Generated by pyupgrade --py36-plus --keep-percent-format.
Now including %d, %i, %u, and multi-line strings.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Use read-only types (List ↦ Sequence, Dict ↦ Mapping, Set ↦
AbstractSet) to guard against accidental mutation of the default
value.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
There seems to have been a confusion between two different uses of the
word “optional”:
• An optional parameter may be omitted and replaced with a default
value.
• An Optional type has None as a possible value.
Sometimes an optional parameter has a default value of None, or None
is otherwise a meaningful value to provide, in which case it makes
sense for the optional parameter to have an Optional type. But in
other cases, optional parameters should not have Optional type. Fix
them.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
To take screenshots, we log in as Desdemonia who doesn't seem
to have permissions to the streams where we capture screenshots.
This commit fixes it by logging in as Iago.
Also added a waitForNavigation until DOMContentLoaded to
avoid flakes that could occur because of us going to another URL
before Iago gets logged in.
Fixes#2665.
Regenerated by tabbott with `lint --fix` after a rebase and change in
parameters.
Note from tabbott: In a few cases, this converts technical debt in the
form of unsorted imports into different technical debt in the form of
our largest files having very long, ugly import sequences at the
start. I expect this change will increase pressure for us to split
those files, which isn't a bad thing.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Automatically generated by the following script, based on the output
of lint with flake8-comma:
import re
import sys
last_filename = None
last_row = None
lines = []
for msg in sys.stdin:
m = re.match(
r"\x1b\[35mflake8 \|\x1b\[0m \x1b\[1;31m(.+):(\d+):(\d+): (\w+)", msg
)
if m:
filename, row_str, col_str, err = m.groups()
row, col = int(row_str), int(col_str)
if filename == last_filename:
assert last_row != row
else:
if last_filename is not None:
with open(last_filename, "w") as f:
f.writelines(lines)
with open(filename) as f:
lines = f.readlines()
last_filename = filename
last_row = row
line = lines[row - 1]
if err in ["C812", "C815"]:
lines[row - 1] = line[: col - 1] + "," + line[col - 1 :]
elif err in ["C819"]:
assert line[col - 2] == ","
lines[row - 1] = line[: col - 2] + line[col - 1 :].lstrip(" ")
if last_filename is not None:
with open(last_filename, "w") as f:
f.writelines(lines)
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
The test_management_commands use in particular was causing pickling
errors when the test failed, because Python 3 filter returns an
iterator, not a list.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Run production suites on Ubuntu Focal.
Added separate success-http-headers files for Focal and Bionic.
Also excluded them from whitespace rules in lint.
memcached 1.5.22 in Ubuntu 20.04 has a bug where it looks for its SASL
configuration at /etc/sasl2/memcached.conf/memcached.conf instead of
/etc/sasl2/memcached.conf.
We already use a workaround for this while applying puppet configurations in
99e71f3786 but for docker builds we used
do memcached hack since we can not use systemd in docker containers.
This was written by Rishi for a very brief purpose a few years ago,
and it doesn't serve much purpose now other than to be a place we
update in code sweeps.
Generated by pyupgrade --py36-plus --keep-percent-format, but with the
NamedTuple changes reverted (see commit
ba7906a3c6, #15132).
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Since we use this option in our docker-zulip project also
so rather than using it as a test suite option we made it
more specific i.e. --build-release-tarball-only.
Some UTF-8 characters (★ for example) are not displayed correctly, with
fonts-liberation. Puppeteer recommends[1] installing fonts-freefont-ttf in
their docs on running Puppeteer in docker.
Provisioning forward is sufficient. There's no need to remove the
new font and replace it with the old font, I think.
[1]: https://github.com/puppeteer/puppeteer/blob/master/docs/troubleshooting.md#running-puppeteer-in-docker
We remove the "GROUP PMs" section that used
to be in the lower right sidebar.
Most of this is straightforward code removal.
A couple quick notes:
- The message fetching code now just
calls `huddle_data.process_loaded_messages`,
which we still need for search suggestions.
We removed `activity.process_loaded_messages`.
- The `huddle_data.process_loaded_messages`
function no longer needs to return `need_resize`.
- In `resize.js` we now just calculate
`res.buddy_list_wrapper_max_height` directly
from `usable_height`.
Fixes#14962
* codecov needs `.coverage` file in the pwd to upload coverage
results.
* concurrency='multiprocessing' forces `.coverage` file to have a
data_suffix, we explicity set it to "" for the file to have no suffix.
Filed issue as https://github.com/nedbat/coveragepy/issues/989
The bug this was working around does not affect our current toolchain,
as confirmed by grepping through the minified output.
(Also, this linter rule only matched calc(x + y) with two arguments
and we were already using calc($far_left_gutter_size + $left_col_size
+ 4px).)
Signed-off-by: Anders Kaseorg <anders@zulip.com>
mock is just a backport of the standard library’s unittest.mock now.
The SAMLAuthBackendTest change is needed because
MagicMock.call_args.args wasn’t introduced until Python
3.8 (https://bugs.python.org/issue21269).
The PROVISION_VERSION bump is skipped because mock is still an
indirect dev requirement via moto.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Previously, we handled this code only in message_list_view.js.
Now we support rendering stream descriptions and some dynamic
elements can be rendered in them, so we extract this new module
and use it in both the places.
This commit adds python code to call javascript_examples.js in its
two supported modes. tools/test-api asserts that the example output
is as expected, whereas the API markdown extension is used to render
these examples in the docs.
Otherwise, if someone tries to create a droplet with a mixed case username,
the existing DNS records (which are always lowercase) would not be deleted.
Also, existing droplets would not be destroyed(--recreate) if there is a mismatch
in case. Unlike DNS records, droplets are created with case sensitive
hostnames.
Plus it's neater to have lowercase hostnames.
As puppeteer tests are similar to casper i.e both being frontend
screen scraping tests, we need to use the same webpack setup as
that of casper's instead of `build_for_most_tests` which is used
mostly for backend related tests.
This sets env variable `CASPER_TESTS` to `1` which does all the
work to make it use the same setup of casper. We would want to
rename that to `PUPPETEER_TESTS` and make changes to do the
same as casper webpack setup when we completely replace casper
with puppeteer.
Zulip's openapi specification in zulip.yaml has various examples
for various schemas. Validate the example with their respective
schemas to ensure that all the examples are schematically correct.
Part of #14100.
We use this new widget in bot settings panels
(personal and org). It lets you re-assign a
bot to a new human user.
Ideally we can improve this code to use
our existing list widgets to make it more
performant for realms with lots of users.
While this functionality to post slow queries to a Zulip stream was
very useful in the early days of Zulip, when there were only a few
hundred accounts, it's long since been useless since (1) the total
request volume on larger Zulip servers run by Zulip developers, and
(2) other server operators don't want real-time notifications of slow
backend queries. The right structure for this is just a log file.
We get rid of the queue and replace it with a "zulip.slow_queries"
logger, which will still log to /var/log/zulip/slow_queries.log for
ease of access to this information and propagate to the other logging
handlers. Reducing the amount of queues is good for lowering zulip's
memory footprint and restart performance, since we run at least one
dedicated queue worker process for each one in most configurations.
We use cloud-config for setting up the SSH keys and executing
some commands. When cloud-config sets the SSH key it doesn't override
the existing keys. So we need to set the SSH keys manually using a command
instead. This means we no longer require cloud config. We can instead
pass a bash script as the user data instead of cloud-config.
I also included a command to set the SSH key of the root.
size_slug represents the plan the droplet should be created on.
Since the new base droplet is created on the cheaper but more
feature rich new plan we have to update the slug_size as well
to take advantage of the cheaper plan.
Previously a YAML syntax error resulted in an
UnhandledPromiseRejectionWarning and a successful exit code; now it
gives a clear message and a failing exit code.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
The major PROVISION_VERSION bump would not be needed, but it was
missing in commit 5ab62a3514 (#14834),
so I’m doing it here.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
We now use the `--streamlined` options for `run-dev.py`
when we use `test_server_running` for `test-api` and
`test-js-with-casper` (and its experimental
replacement, `test-js-with-puppeteer`).
This means we don't slow anything down with
processes like thumbor, process_fts_updates, etc.,
which aren't meaningfully exercised by these tests.
We may eventually want some tests to meaningfully
exercise those processes, and when that day comes,
we will need to add an extra argument to
`test_server_running`, probably, but until then,
we just always set `--streamlined` in that codepath.
There is actually a tool called `./tools/test-run-dev`
that we run in CI, and it will use the full mode.
It just doesn't verify much stuff--it mostly polls
the server without testing specific features.
This seems to save about 1s of the startup time on a system I use
(~10.6s -> ~9.7s).
For basic testing (either manual or automated), we
generally only need the server and tornado running.
Obviously, it's nice to test the complete system,
but if you're on a slow PC, the overhead can be
annoying.
Note that we don't launch any of these processes
in `--streamlined` mode:
process_queue
process_fts_updates
deliver_scheduled_messages
thumbor
And then by not launching process_queue, we avoid
several child processes.
Basic functionality like sending messages will
still work here.
The streamlined mode may be helpful in debugging
our generally slow server startup time. Obviously,
some of the problem with startup is the auxiliary
processes here, but removing them as a variable
could help us focus on getting the core stuff fast.
Note that we still have the webpack watcher running
in streamlined mode.
For the particular case of thumbor, note that we
modify the proxy server to explicitly print and
return an error if we get a `/thumbor/*` request.
We clean up the code related to launching
processes here.
We extract:
server_processes
We also extract these helper for webpack
stuff:
do_one_time_webpack_compile
start_webpack_watcher
And then we move the code to actually launch
them lexically within the file (so as not to
be obscured by various function definitions).
Here is the new output for displaying ports:
Zulip services will listen on ports:
9991: web proxy
9992: Django
9993: Tornado
9994: webpack
9995: Thumbor
Note to Vagrant users: Only the proxy port (9991) is exposed.
I tone down the yellow for the Vagrant warning, and I show
the web proxy in cyan to emphasize it.
I also extracted the code into a function, and I don't call
that function until after `app.listen()`. (The users probably
won't notice much difference in the timing of this message, but
the message won't show if the `listen` step fails for some
reason, which I think is what we want here.)
We remove the import-tools code that was plunked
right into the middle of our command line
arguments.
Then we add a local var called `DESCRIPTION` to
fix some ugly code formatting, and we stop with the
unnecessary `r` prefix to the multi-line string.
This does not rely on the desktop app being able to register for the
zulip:// scheme (which is problematic with, for example, the AppImage
format).
It also is a better interface for managing changes to the system,
since the implementation exists almost entirely in the server/webapp
project.
This provides a smoother user experience, where the user doesn't need
to do the paste step, when combined with
https://github.com/zulip/zulip-desktop/pull/943.
Fixes#13613.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This will give help up write new digest only if the db rebuild
succeeds. We were relying on the caller to
be successful in building db, this was hacky and unreliable.
We write new db digest once the caller succeeds, this ensures
that we write new digest after every successful attempt.
This fixes the anomality we were facing that Databases were rebuild
on the 2nd provision attempt with no changes to files or migrations.
This was happening because we didn't write a new digest for db
after the first provision (The case of DB didn't exist).
During the 1st provision, we check the template_status() of
Database both Dev and Test, but database_exists() of Databases
obviously returned false, and we rebuild the database,
but forgot to write_new_digest and hence the anomaly in the
second provision explained above.
Yes, it's slightly janky to create an
argparse.Namespace object like this, but it
saves us from shelling out to a script whose
only real value-add is parsing a single
`threshold_days` argument.
This saves about 130ms for a no-op provision.
We now just have two modes for setting up a dev/test
database. This makes it easy to see these things
side-by-side, when you're trying to understand how
the two different databases get built:
# dev:
USERNAME=zulip
DBNAME=zulip
STATUS_FILE_NAME=migration_status_dev
# test:
USERNAME=zulip_test
DBNAME=zulip_test
STATUS_FILE_NAME=migration_status_test
And then we make it more explicit the things that
are common between dev and test (which are
important things to understand when troubleshooting
provision-related glitches):
SEARCH_PATH=zulip,public
PASSWORD=$("$(dirname "$0")/../../scripts/get-django-setting" LOCAL_DATABASE_PASSWORD)
DBNAME_BASE=${DBNAME}_base
We lose some "generality" here, but passing in arbitrary
combinations of username/dbname/status_file to the script
would cause chaos for our digest checks, and all the different
template/base databases could cause confusion too.
We now prevent these variations:
* <hr/>
* <hr />
* <br/>
* <br />
We could enforce similar consistency for other void
tags, if we wished, but these two are particularly
prevalent.
Instead of figuring out the image path from the integration name in the
puppeteer script, we do it in the `generate-integration-docs-screenshot`
script and pass it as an argument to `message-screenshot.js`.
Since in travis we don't have root access so we used to add different
srv path. As now we shifted our production suites to Circle CI
we don't need that code so removed it.
Also we used a hacky code in commit-lint-message for travis which is
now of no use.
The test-all tool is useful if you want to be really sure everything
passes, but it's almost always better to run a smaller subsuite when
doing active development, and just run test-all (or CI) at the end.
Our docs already covered this issue well, but this presents the idea
in a place one is definitely looking at when running test-all.
Previously, we added support for 'none', 'plain' and 'noop' and a
function `lang = remap_language(lang)`. This also had the potential
to encourage adding more remappings- something that we deliberatly
want to keep to a minimum.
For context, Anders K doesn't want us to keep any remapping (only
keeping 'text' which is the default no-op lexer that pygments has)
and Tim wants to keep 'plain' and 'text'. We should only document
and advertise 'text'.
In Django 2.1, the preferred way to express a nullable BooleanField
changed from NullBooleanField to passing null=True to BooleanField.
This updates our codebase to use the preferred API. Tweaked by
tabbott to update the linter rules.
The migration is a noop for Django accounting only.
Part of #11341.
This allows to run scripts between extraction and install
process.
It will be used to restore npm caches for production install jobs.
We extract the tarball in the working directory so that yarn.lock and
package.json are available to restore cache.
(And also so the path is deterministic).
Now that we've cleaned up this tool's output, there's no reason to use
an awkward mechanism to hide its output; we can just print it out like
a normal program.
Fixes#14644; resolves#14701.
We now forbid tags of the form `<foo ... />` in most
places, and we also forbid it even for several void
tags.
We make exceptions for tags that are already formatted
in two different ways in our codebase. This is mostly
svg tags, plus these common cases:
- br
- hr
- img
- input
It would be nice to lock down a convention for these,
even though the HTML specification is unopinionated
on these. We'll probably want to stay flexible for
svg tags, since they are sometimes copy/pasted from
other sources (although it's probably rare enough for
them that we can tolerate just doing minor edits as
needed).
If folks put something like '<br/>' in the HTML,
we would think the tag's name was "br/" instead
of "br". I think we were assuming most folks
would write either "<br>" or <br />".
ASIDE:
We should probably have a consistent
preference among these styles:
* <br>
* <br/>
* <br />
I prefer the first.
Runs a test at the end of tools/ci/backend to check if any untracked
files have been created during this ci tests. Exits with error code 1
if untracked files are found, otherwise exits successfully.
Fixes#14691.
We recently changed our droplet setup such that their
host names no longer include zulipdev.org. This caused
a few things to break.
The particular symptom that this commit fixes is that
we were trying to server static assets from
showell:9991 instead of showell.zulipdev.org:9991,
which meant that you couldn't use the app locally.
(The server would start, but the site's pretty unusable
without static assets.)
Now we rely 100% on `dev_settings.py` to set
`EXTERNAL_HOST` for any droplet users who don't set
that var in their own environment. That allows us to
remove some essentially duplicate code in `run-dev.py`.
We also set `IS_DEV_DROPLET` explicitly, so that other
code doesn't have to make inferences or duplicate
logic to detemine whether we're a droplet or not.
And then in `settings.py` we use `IS_DEV_DROPLET` to
know that we can use a prod-like method of calculating
`STATIC_URL`, instead of hard coding `localhost`.
We may want to iterate on this further--this was
sort of a quick fix to get droplets functional again.
It's possible we can re-configure droplets to have
folks get reasonable `EXTERNAL_HOST` settings in their
bash profiles, or something like that, although that
may have its own tradeoffs.
Avatar images for bots used by the tool to generate integration
documentation screenshots are pre-generated and committed to the repository.
The `generate-integration-docs-screenshot` tool now uses these images,
instead of trying to create these avatar images on the fly.
Also, deleted the unused `create_png_from_svg` function.
We now have helpers for the two places where
we create databases.
There was already one helper in place, and
I gave it a more concrete name, to match
its actual database name in postgres.
So `generate-fixtures` only ever did 9 lines of code (really
3 lines of actual code) in its normal mode of operation.
But it was cluttered with lots of stuff that really only
happened when you called it with the `-force` option, which
was only invoked by `rebuild-test-database`.
Now we inline most of the code into `rebuild-test-database`.
And now `generate-fixtures` is simple (and doesn't support
a `-force` flag.
We remove the `generate_fixtures` option here mostly
for simplicity, but in particular to facilitate
an upcoming commit to simplify the job of
`generate-fixtures` (and remove its `--force` option).
The command line option here for `test-backend`
was really calling `generate_fixtures --force`,
which we're about to rename `tools/rebuild-test-database`.
The `test-backend` tools is already smart about catching
up on migrations, so we generally don't need to tell it
to repair the database.
And if the database does get corrupt, you can just do
it directly with `tools/rebuild-test-database`.
This eliminates the `use_force` flag in
`update_test_databases_if_required`, which was easy
to confuse with `rebuild_test_database`.
The other caller wasn't using `use_force`.
The new tools now have more concise, more parallel names:
- rebuild-dev-database
- rebuild-test-database
The actual implementations are still pretty different:
rebuild-dev-database:
mostly delegates to 5 management scripts
rebuild-test-database:
is a very thin wrapper for generate-fixtures
We'll try to clean that up a bit soon.
This tool was part of a very ad hoc investigation
during 2017 into our JS dep dependencies.
It's very out of date, and it has a non-trivial
maintenance cost, as these type of tools seem
to come up in every code sweep.
Commit 35577a1f66 (#9406) moved the
OpenAPI definitions to zerver/openapi, and this script was not
updated, so it has been validating nothing for two years.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Restart postgres service if provision is called in production test suite.
This is required because terminate-psql-sessions script (used
in tools/ci/setup-production) throws error if postgres service is not running.
Restart rabbitmq service if provision is called in production test suite.
This is done to start the node as Circle CI don't start services on installation.
Removed memcached restart as flush-memcached script (which is furthur
used in tools/ci/production) throws UNKNOWN READ FAILURE if memcached is restarted
in development.
Since now we want to use production suites on Circle CI so there
is no need to set TRAVIS in env while running scripts.
CIRCLECI is set default in the enviroment of Circle CI builds
so we can use it directly.
Also Travis CI had rabbitmq-server installed so we had to add workaround
in install script to avoid the error. That workaround is removed.
Used postgres 10 inplace of postgres 9.5 as it is used in Bionic.
Upgraded nginx version in success-http-headers which is in CircleCI
Bionic enviroment.
This seems to be a complicated wrapper around `sed`.
It's not used anywhere or documented anywhere. I
assume it was used in some code sweep or something.
See cec45b0ae5
I built this in 2013 to help me quickly build
some sample data for our very early node tests,
I believe.
Ever since then it's just been changed for various
code sweeps. Also, if we wanted to resurrect this
idea for some reason, we now have the template
parser that was written since then.
This tool was last modified in 2016, has no
documentation, and requires `dot` to run, so
I think we can remove it to reduce some clutter
in the tools directory.
This tool is hackish and incomplete, but it's still
worth looking at if somebody wants to hunt down
obsolete CSS. There are probably better ways to
tackle this problem, so we should eventually just
remove this tool, but it's pretty low maintenance.
Current output looks like this:
$ ./tools/find-unused-css
actions_hovered
actions_link
bookend_tr
btn-skip
company-name
flatpickr-months
label_for_text
loading_more_messages_indicator_box
loading_more_messages_indicator_box_container
logoimage
messages-collapse
messages-expand
numInputWrapper
page_loading_indicator_box
page_loading_indicator_box_container
skinny-user-gravatar
sp-input
summary_colorblock
summary_row
summary_row_private_message
tutorial-done-button
Else for each retry the duplicate commits would be removed
again and again from the contributor's zulip/zulip commits.
This is a bug in the original commit
6bed6ccdcf that added the
functionality to remove duplicate commits.
We now have two functions related to digests
for processes:
is_digest_obsolete
write_digest_file
In most cases we now **wait** to write the
digest file until after we've successfully
run a process with its new inputs.
In one place, for database migrations, we
continue to write the digest optimistically.
We'll want to fix this, but it requires a
little more code cleanup.
Here is the typical sequence of events:
NEVER RUN -
is_digest_obsolete returns True
quickly (we don't compute a hash)
write_digest_file does a write (duh)
AFTER NO CHANGES -
is_digest_obsolete returns False
after reading one file for old
hash and multiple files to compute
hash
most callers skip write_digest_file
(no files are changed)
AFTER SOME CHANGES -
is_digest_obsolete returns False
after doing full checks
most callers call write_digest_file
*after* running a process
There's no real reason to do the lazy import any
more, as we use this unconditionally inside `main`
(indirectly), and `provision_inner` runs after we
have set up the venv.
I make these all functions for consistency,
and in particular I want to continue to avoid
`glob.glob` calls until we are actually
computing hashes.
This is mostly a prep to allow us to do
hashing in two separate places:
- check hashes
- update hashes
We would only update hashes **after** running
processes anew.
For `provision_inner` I considered using a
class to put the three path-related helpers
into a mini namespace, but it felt too heavy.
It wouldn't be completely implausible here
to extract something like a JSON config
file that has a list of globs for each
process that we do path-hashing for, but I
want to clean up other stuff first.
Importing cairosvg in fails in production, because `libgtk-3-dev` is not
available. This commit moves the import for `cairosvg` into the function
where it is used. This code will soon be moved into a separate script that
will not be run on production.
This guarantees that we don’t accidentally upgrade one without the
other, which could happen for example due to different third-party
version constraints between the two.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
We no longer need to maintain duplicate code
related to where we set up the emoji
cache directory.
And we no longer need two extra steps for
people doing advanced (i.e. manual) setup.
There was no clear benefit to having provision
build the cache directory for `build_emoji`,
when it was easy to make `build_emoji` more
self-sufficient. The `build_emoji` tool
was already importing the library that has
`run_as_root`, and it was already responsible
for 99% of the create-directory kind of tasks.
(We always call `build_emoji` unconditionally from
`provision`, so there's no rationale in terms
of avoiding startup time or something.)
ASIDE:
Its not completely clear to me why we need
to put this directory in "/srv", instead of
somewhere more local (like we already do for
Travis), but maybe it's just to be like
its siblings in "/srv":
node_modules
yarn.lock
zulip-emoji-cache
zulip-npm-cache
zulip-py3-venv
zulip-thumbor-venv
zulip-venv-cache
zulip-yarn
I guess the caches that we keep in var are
dev-only, although I think some of what's under
`zulip-emoji-cache` is also dev-only in nature?
./var/webpack-cache
./var/mypy-cache
In `docs/subsystems/emoji.md` we say this:
```
The `build_emoji` tool generates the set of files under
`static/generated/emoji` (or really, it generates the
`/srv/zulip-emoji-cache/<sha1>/emoji` tree, and
`static/generated/emoji` is a symlink to that tree;we do this in
order to cache old versions to make provisioning and production
deployments super fast in the common case that we haven't changed the
emoji tooling). [...]
```
I don't really understand that rationale for the development
case, since `static/generated` is as much ignored by `git` as
'/srv' is, without the complications of needing `sudo` to create it.
And in production, I'm not sure how much time we're really saving,
as it takes me about 1.4s to fully rebuild the cache in dev, not to
mention we're taking on upgrade risk by sharing files between versions.
So, `source_emoji_dump` is not the greatest variable
name, but at least we now define it relative to its
parent instead of the `.success-stamp` file. (And
then `success_stamp` is just another join.)
If the directory `templates/zerver/emails/compiled/`
is missing, then we need to run `inline_email_css`
again.
This can happen if somebody gets overzealous about
cleaning untracked files.
This is more encapsulated and more efficient.
In the cases where `is_force` is `True` or
`pygments_data.json` is missing, we now avoid
the unnecessary step of importing `pygments`, at
least up front.
(Of course, we probably import that once we generate
the artifacts.)
If somebody is having issues with provision, it's
plausible they'll do something like `git clean -fX`
to clean up old artifacts of earlier provision runs,
as part of debugging things.
We defend against this by detecting the most obvious
symptom as cheaply as possible.
I remove `is_force` from `file_or_package_hash_updated`
and modernize its mypy annotations.
If `is_force` is `True`, we just now run the thing
we want to force-run without having to call
`file_or_package_hash_updated` to expensively
and riskily return `True`.
Another nice outcome of this change is that if
`file_or_package_hash_updated` returns `True`,
you can know that the file or package has
indeed been updated.
For the case of `build_pygments_data` we also
skip an `os.path.exists` check when `is_force`
is `True`.
We will short-circuit more logic in the next
few commits, as well as cleaning up some of
the long/wrapper lines in the `if` statements.
We change the message for skipping RabbitMQ
configuration to match nearby messages:
No need to run `tools/setup/build_pygments_data`.
No need to run `scripts/setup/inline_email_css.py`.
No need to run `scripts/setup/configure-rabbitmq.
No need to regenerate the dev DB.
No need to regenerate the test DB.
No need to run `manage.py compilemessages`.
For upgrade-zulip-from-git to work, we need to be able to run
update-prod-static on production systems, which means provision code
like this cairosvg logic needs to be there for now.
When creating a webhook integration or creating a new one, it is a pain to
create or update the screenshots in the documentation. This commit adds a
tool that can trigger a sample notification for the webhook using a fixture,
that is likely already written for the tests.
Currently, the developer needs to take a screenshot manually, but this could
be automated using puppeteer or something like that.
Also, the tool does not support webhooks with basic auth, and only supports
webhooks that use json fixtures. These can be fixed in subsequent commits.
We figure out the dev host using the same logic as
dev_settings.py, so that we don't use wrong things
like 127.0.0.1 for droplet users.
And we display the link in cyan.
Generated by `pyupgrade --py3-plus --keep-percent-format` on all our
Python code except `zthumbor` and `zulip-ec2-configure-interfaces`,
followed by manual indentation fixes.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
If a request fails the tool sleeps for some time before making
further requests. The sleep time is a random number between
0 and 2^failures capped at 64 seconds. More details about the
algorithm can be found at https://chat.zulip.org/#narrow/stream/
92-learning/topic/exponential.20backoff.20--.20with.20jitter
This is a prep commit for generating /team page data
using cron job. zerver/tests directory is not present in
production installation. So moving the file from the directory
tests to tools.
As described in the commit that added this function, this fixes one
quite annoying bug and one at least in-principle bug:
* On Windows, the simple version (lacking `git update-index
--refresh`) routinely gives false positives, making the tools
that rely on it basically unusable.
* If you have uncommitted changes in the index but manage to have
the worktree nevevertheless match HEAD, the simple version will
give a false negative and we'd blow away those changes.
This is verbatim from Git upstream, at an older version. (The one
change since then is to add localization for the messages like "You
have unstaged changes" -- which complicates the code, is important and
worth it for Git itself, but for our tools we can do without.)
This function will replace our use of `git diff-index --quiet HEAD`
in several scripts. The key differences in behavior are:
* The `git update-index --refresh`. Without this, on Windows
apparently `git diff-index` routinely (but not all the time!)
reports that tons of files have changed. See report:
https://chat.zulip.org/#narrow/stream/9-issues/topic/.2E.2Ftools.2Ffetch-pull-request.20issue/near/834435
* Instead of one command comparing the worktree to HEAD, we
separately compare the worktree to the index and the index to
HEAD, and abort if either diff is nonempty. This one is obvious,
but rather an edge case (it matters only if you've managed to
make the worktree and HEAD agree while the index has some
changes), and the extra code is annoying if written out in every
script that needs it. But that's what a subroutine is for. :-)
We'll make a few tweaks before actually switching to use this.
The Git commands we're invoking to do the real work are useful to
print, for transparency to see what's happening and that there's no
magic here.
The boring shell stuff like `remote=${2:-"upstream"}` is not so
helpful, and nor is the rather arcane and in any case read-only
command `git diff-index --quiet HEAD`. Those only add noise that
obscures the interesting parts. So, move the `set -x` down to when
we're done with the boring preparatory stuff and ready to perform
the commands that do the work.
Add sgrep (sgrep.dev) to tooling and include simple rule as
proof of concept. Included rule detects use of old django render
function.
Also added a rule that looks for if-else statements where both
code paths are identical.
While we could fix this issue by changing the markdown processor,
doing so is not a robust solution, because even a momentary bug in the
markdown processor could allow cached messages that do not follow our
security policy.
This change ensures that even if our markdown processor has bugs that
result in rendered content that does not properly follow our policy of
using rel="noopener noreferrer" on links, we'll still do something
reasonable.
Co-authored-by: Tim Abbott <tabbott@zulipchat.com>
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Folks can have issues connecting to Casper
as zulipdev.com when they are not connected to
the internet or just have a bad connection, since
the DNS record is on the internet. Folks can
work around this by just creating an /etc/hosts
entry for zulipdev.com, but people don't always
know.
This fix moves the symptom slightly earlier in
the process--we don't advertise that the server
is "up" if you can't actually connect to it as
"zulipdev.com".
In the past it has blocked Python library security updates with overly
strict version bounds, and we don’t use it as a library, only as a
binary.
Skip the PROVISION_VERSION bump because we can use the tx binary from
either location.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This reverts commit 36a8e61e67 (#13934).
The Django 2.2 autoreloader works by forking into a child process that
exits with status 3 when a file changes, and a parent process that
restarts the child when it exits with status 3. Setting this
environment variable had the effect of pretending we were already the
child process, without a parent process to restart it. Therefore,
changing any code used by the queue processor caused it to exit rather
than restart.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
python-dev will be depreciated in Focal but can be used as python2-dev
so removed it from common dockerfile.template and added it
as an extra package in .circleci/config.yml.
Previously, we only did apt updates when our sources.list files or
keys changed, which could result in provisioning errors for
development systems that don't routinely update their apt cache
(probably including ~all Vagrant environments).
cgi.escape is deprecated in python3.2 and removed in python3.8.
This function was unsafe because quote is false by default, hence
removed and replaced with a safer html.escape.
Used get_venv_dependencies function to return the correct dependencies
for RHEL, Centos, Fedora rather than importing them as separate
COMMON_YUM_DEPENDENCIES in provision and create-production-venv.
We now restrict emails on the zulip realm, and now
`email` and `delivery_email` will be different for
users.
This change should make it more likely to catch
errors where we leak delivery emails or use the
wrong field for lookups.
Tests for these links often result in rate-limiting from GitHub,
leading to the builds failing in Circle CI. We temporarily mark
github.com/zulip links as external to keep the builds passing.
Added a get_venv_dependencies() function in setup_venv.py which
returns VENV_DEPENDENCIES according to the vendor and os_version.
The reason for adding this function was because python-dev will be
depreciated in Focal but can be used as python2-dev so when adding
support for Focal VENV_DEPENDENCIES should to be os_version dependent.
There were two problems with the previous code-
1) The code glob.glob("scripts/lib/build-") should be
glob.glob("scripts/lib/build-*) otherwise it would always return [].
2) The part of the code where we included scripts/lib/build-* for sha1 sum
check would only run when debian is not in os_families(). This wasn't
correct as we could have a situation where we have to build pgroonga
from source even in case of debian and so we need to improve the
condition on it.
Now since we only have build-pgroonga there its better to just directly hash
its content with the condition of BUILD_PGROONGA_FROM_SOURCE.
This should fix spurious failures, where test-run-dev would occasionally
freeze. What exactly about these changes was causing that is still to
be potentially investigated. This is merely meant as a fix to the
failures.
This reverts commit 19429c3ad7.
We plan to use these records to check and record the schema of Zulip's
events for the purposes of API documentation.
Based on an original messier commit by tabbott.
In theory, a nicer version of this would be able to work directly off
the mypy type system, but this will be good enough for our use case.
Before this test, we were validating the behavior
of `i18next`, but we weren't validating our light
layer that sits on top of `i18next`, which currently
resides in the slightly misnamed `translations.js`
file.
The translations module is now so small that I'll
just quote it verbatim here:
import i18next from 'i18next';
i18next.init({
lng: 'lang',
resources: {
lang: {
translation: page_params.translation_data,
},
},
nsSeparator: false,
keySeparator: false,
interpolation: {
prefix: "__",
suffix: "__",
},
returnEmptyString: false, // Empty string is not a valid translation.
});
window.i18n = i18next;
We now just do `zrequire('translations')` to initialize
the `i18next` library, which allows us to have simpler
test setup and to actually exercise the above call to
`i18next.init`.
This change now gives us 100% line coverage of `translations.js`,
which of course isn't that hard to acheive (see above).
isort 5 knows not to reorder imports across function calls, so this
will stop isort from breaking our code.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This gives them cache-compatible URLs, and also avoids some extra
copies of the sprite sheet images.
Comments on the Octopus emoji added by tabbott.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Now the caller simply imports the debug ‘require’ function as a
module, deciding for itself how to expose it and with what name (in
our case, we expose it as ‘require’ with expose-loader). Also, remove
a stray console.log.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This adds a global require() function that makes JS modules accessible
to the browser console without adding them to the global window
object:
» const typeahead = require("./static/shared/js/typeahead");
» typeahead.popular_emojis
Array(6) [ "1f44d", "1f389", "1f642", "2764", "1f6e0", "1f419" ]
The list of known modules is exposed via the keys of require.ids
object.
This will allow us to migrate more modules to ES6 without losing
access to this debugging functionality.
I’ll probably upload this plugin to NPM at some point, but I figured
I’ll let it bake in-tree first.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This moves some code from settings_display.js
into the new module settings_config.js.
Extracting this module breaks some dependencies
on settings_display.js (which has some annoying
transitive dependencies, including jQuery).
In particular this isolates stream_data from
from settings_display.js.
Two of the three structures that we moved here
weren't even directly used by settings_display.js,
since we do a lot of rendering in the modules
admin.js and setting.js.
We make get_all_display_settings() a function
to avoid a require-time dependency on page_params.
Breaking the dependencies simplifies a few
node tests.
Most of the node test complexity came from the
following commit in March 2019:
5a130097bf
The commit itself seems harmless enough, but
dependencies can have a somewhat "viral" nature,
where making stream_data depend on settings_display
caused us to modify four different node tests.
This cleans up a few things:
- just yield values so we don't have to do
tedious max logic
- use values() instead of items() for
skin_variations loop
In the ideal world the emoji.json would reduce this
code to `get_square_size = lambda data: data['square_size']`,
but I don't think we can get the square size explicitly.
This commit changes the calculation of the
background-size parameter that we use to
render emojis from sprite sheets.
In particular, it now makes the parameter
match the sizes of our latest sprite
sheets from Twitter/Google.
This should fix the geometry aspect of #13959,
but we also need to fix some issues with the
cache being sticky.
There is also some minor cleanup:
- Remove obsolete -moz/-webkit CSS.
- Remove needless precision in percentages.
- Fix the transposed nrows/ncols names.
- Add extensive commenting.
Finally, we add a minor bump to the provision
number. This commit should be merged in the
same series as the other fix for this issue,
which will probably have a major bump, and we'll
need to rebase this appropriately.
While it's a bit of extra complexity to do this check, which I'm not
excited about, we've had multiple folks spend significant time being
confused rebasing past d7d8632525 into
deleting `pygments_data.json`, with provision not rebuilding it, so
this seems worth merging as a transitional fix even if we decide to
remove it in 2 months.
Without calling cov.erase() the data file seems to persist and even
pollute future test runs if not removed. Registering an atexit handler
seems like a good, and reasonably clean way to ensure the cleanup
happens.
Fixes#13933.
This allows us to collect coverage for Handlebars templates, and also
improves the readability of Handlebars-related stack traces.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
In Django 2.2 the autoreload system has changed.
DJANGO_AUTORELOAD_ENV env variable should be set when calling code
that'll use the autoreloader. Otherwise there's some kind of race
condition in the autoreload code when SIGINT is sent, where
restart_with_reloader() (called only if the env variable isn't set)
has the subprocess module calling p.kill() on a process that's already
exited, raising ProcessLookupError and printing an ugly traceback. This
causes non-deterministic test-run-dev failures.
We used to have a block of code doing this just in the presence
endpoint because that's where we'd had error-handling problems with it
not being present, but it seems more correct for it to run
unconditionally on all HTTP requests.
This requires adding a dependency of channel on reload_state, which we
record in the webpack configuration for now.
This should return us to a situation where we won't get blueslip
browser error reporting for users created while a device was offline
just before it reloads.
1) Created a new class `DatabaseType` and access its objects inside
`template_database_status()` instead of sending five arguments with
default values.
2) Made `check_files` and `setting_name` local variables instead of
function parameters since they had same value(None) for every call.
Fixes#13845.
webpack optimizes JSON modules using JSON.parse("{…}"), which is
faster than the normal JavaScript parser.
Update the backend to use emoji_codes.json too instead of the three
separate JSON files.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
I believe we can remove these and rely on
other parts of our testing/code-review
to ensure template quality.
These tests never really exercised our
app code, as evidenced by us not regressing
any of the 100%-line-coverage files.
We have a couple other ways that we verify
the correct format of the templates:
- webpack (can they compile?)
- check-templates (are they nicely indented?)
For deep testing, we have Casper, which
exercises most of our most important templates
in some meaningful way.
I think it's pretty rare that we get bugs
now that are directly caused by bad templates,
and an even smaller subset of them would
have been caught by the node tests.
If that trend changes in the future, I would prefer to
just do something "greenfield" to address
any common problems rather than resurrect
this code, but we could always resurrect it
from git.
The template node tests did check a little bit of
detail about which fields are there, but not
in an integrated way, so that aspect of the tests
wasn't very useful either.
This adds Ubuntu 19.10 as a valid provisioning target.
The release test in setup-apt-repo was changed from a list of values to
a regex check for brevity.
Every CLI program should have a usage message.
Also add a mention in the `push-to-pull-request` usage message of
its participation in the `refs/remotes/pr/` pseudo-remote feature.
This gives us the right behavior when using the `url.*.insteadOf`
mechanism for aliases in Git remote URLs. For example, if
one's ~/.gitconfig has:
[url "git@github.com:"]
insteadOf = gh:
then `git remote add upstream gh:zulip/zulip` will work great, as
the nice, short, mnemonic `gh:` prefix gets expanded to the more
finicky `git@github.com:`. I use just such a prefix routinely.
But the feature does require that scripts go through the right
abstractions. In particular `git remote get-url`, since Git 2.7
(from 2016), exists for exactly this reason. A plain `git config`
command bypasses the expansion, getting the verbatim `gh:...`
version, which doesn't work.
So, switch to that.
As a bonus, we get to behave correctly if for some reason the user
has configured a push URL distinct from the fetch URL for this
remote, just by adding `--push`. With `git config`, we'd have had
to manually implement the fallback from `remote.upstream.pushUrl` to
`remote.upstream.url` in order to properly handle that case.
We prefer this to internal_send_message().
We are trying to deprecate `internal_send_message`,
which has extra moving parts related to
`extract_recipients` and `Addressee.legacy_build`.
There are two chunks of code that I touch here
that look pretty similar, but I'm not quite
sure they're worth de-duplicating, since they
use different topics and different message
content.
Instead of having `notify_new_user` delegate
all the heavy lifting to `send_signup_message`,
we just rename `send_signup_message` to be
`notify_new_user` and remove the one-line
wrapper.
We remove a lot of obsolete complexity:
- `internal` was no longer ever set to True
by real code, so we kill it off as well
as well as killing off the internal_blurb code
and the now-obsolete test
- the `sender` parameter was actually an
email, not a UserProfile, but I think
that got past mypy due to the caller
passing in something from settings.py
- we were only passing in NOTIFICATION_BOT
for the sender, so we just hard code
that now
- we eliminate the verbose
`admin_realm_signup_notifications_stream`
parameter and just hard code it to
"signups"
- we weren't using the optional realm
parameter
There's also a long ugly comment in
`get_recipient_info` related to this code
that I amended for now.
We should try to take action in a subsequent
commit.
We now have 100% line coverage on 71 JS files.
This is thanks to about 150 people who have
contributed code to frontend/node_tests.
And then 126 files are still short of 100% line
coverage.
We now enforce line coverage with a set called
EXEMPT_FILES, which are the files for which
we do NOT expect to have 100% coverage.
Using an exemption list makes it so that adding
a new JS file to the project without 100% line
coverage will cause the build to fail. This will
encourage folks to be intentional about their
lack of test coverage.
If a file that had 100% coverage somehow regressed
to 0% coverage, we would report an error to the
console, but we weren't treating it as an actual
failure.
We've probably always had this bug, but it probably
rarely was an issue, since devs might have seen
the error locally, or hopefully whatever crazy
thing you did to totally remove coverage would
have had other symptoms.
If this was intentional, I suspect it might have
had something to do with wanting to get coverage
reports when you just run individual tests. But
a while back we changed it so that when you run
individual tests, we don't do the line coverage
enforcement.
My upstream remote is named origin, so commit-message-lint was always
complaining at me. Detect the right remote name from the output of
`git remote -v`.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
We now use vdom-ish techniques to track the
list items for the pm list. When we go to update
the list, we only re-render nodes whose data
has changed, with two exceptions:
- Obviously, the first time we do a full render.
- If the keys for the items have changed (i.e.
a new node has come in or the order has changed),
we just re-render the whole list.
If the keys are the same since the last re-render, we
only re-render individual items if their data has
changed.
Most of the new code is in these two modules:
- pm_list_dom.js
- vdom.js
We remove all of the code in pm_list.js that is
related to updating DOM with unread counts.
For presence updates, we are now *never*
re-rendering the whole list, since presence
updates only change individual line items and
don't affect the keys. Instead, we just update
any changed elements in place.
The main thing that makes this all work is the
`update` method in `vdom`, which is totally generic
and essentially does a few simple jobs:
- detect if keys are different
- just render the whole ul as needed
- for items that change, do the appropriate
jQuery to update the item in place
Note that this code seems to play nice with simplebar.
Also, this code continues to use templates to render
the individual list items.
FWIW this code isn't radically different than list_render,
but it's got some key differences:
- There are fewer bells and whistles in this code.
Some of the stuff that list_render does is overkill
for the PM list.
- This code detects data changes.
Note that the vdom scheme is agnostic about templates;
it simply requires the child nodes to provide a render
method. (This is similar to list_render, which is also
technically agnostic about rendering, but which also
does use templates in most cases.)
These fixes are somewhat related to #13605, but we
haven't gotten a solid repro on that issue, and
the scrolling issues there may be orthogonal to the
redraws. But having fewer moving parts here should
help, and we won't get the rug pulled out from under
us on every presence update.
There are two possible extensions to this that are
somewhat overlapping in nature, but can be done
one a time.
* We can do a deeper vdom approach here that
gets us away from templates, and just have
nodes write to an AST. I have this on another
branch, but it might be overkill.
* We can avoid some redraws by detecting where
keys are moving up and down. I'm not completely
sure we need it for the PM list.
If this gets merged, we may want to try similar
things for the stream list, which also does a fairly
complicated mixture of big-hammer re-renders and
surgical updates-in-place (with custom code).
BTW we have 100% line coverage for vdom.js.
This legacy cross-realm bot hasn't been used in several years, as far
as I know. If we wanted to re-introduce it, I'd want to implement it
as an embedded bot using those common APIs, rather than the totally
custom hacky code used for it that involves unnecessary queue workers
and similar details.
Fixes#13533.
responses is an module analogous to httpretty for mocking external
URLs, with a very similar interface (potentially cleaner in that it
makes use of context managers).
The most important (in the moment) problem with httpretty is that it
breaks the ability to use redis in parts of code where httpretty is
enabled. From more research, the module in general has tendency to
have various troublesome bugs with breaking URLs that it shouldn't be
affecting, caused by it working at the socket interface layer. While
those issues could be fixed, responses seems to be less buggy (based
on both third-party reports like ckan/ckan#4755 and our own experience
in removing workarounds for bugs in httpretty) and is more actively
maintained.
In this commit, we basically match any kinda of jinja2 start tag,
no matter its special kind (eg. jinja2_whitespace_stripped_start)
to any kinda jinja2 end tag (eg. jinja2_whitespace_stripped_end)
Idea is special operators like `-` do not change the meaning of
inline tag and thus matching shouldn't depend upon this.
Zulip has had a small use of WebSockets (specifically, for the code
path of sending messages, via the webapp only) since ~2013. We
originally added this use of WebSockets in the hope that the latency
benefits of doing so would allow us to avoid implementing a markdown
local echo; they were not. Further, HTTP/2 may have eliminated the
latency difference we hoped to exploit by using WebSockets in any
case.
While we’d originally imagined using WebSockets for other endpoints,
there was never a good justification for moving more components to the
WebSockets system.
This WebSockets code path had a lot of downsides/complexity,
including:
* The messy hack involving constructing an emulated request object to
hook into doing Django requests.
* The `message_senders` queue processor system, which increases RAM
needs and must be provisioned independently from the rest of the
server).
* A duplicate check_send_receive_time Nagios test specific to
WebSockets.
* The requirement for users to have their firewalls/NATs allow
WebSocket connections, and a setting to disable them for networks
where WebSockets don’t work.
* Dependencies on the SockJS family of libraries, which has at times
been poorly maintained, and periodically throws random JavaScript
exceptions in our production environments without a deep enough
traceback to effectively investigate.
* A total of about 1600 lines of our code related to the feature.
* Increased load on the Tornado system, especially around a Zulip
server restart, and especially for large installations like
zulipchat.com, resulting in extra delay before messages can be sent
again.
As detailed in
https://github.com/zulip/zulip/pull/12862#issuecomment-536152397, it
appears that removing WebSockets moderately increases the time it
takes for the `send_message` API query to return from the server, but
does not significantly change the time between when a message is sent
and when it is received by clients. We don’t understand the reason
for that change (suggesting the possibility of a measurement error),
and even if it is a real change, we consider that potential small
latency regression to be acceptable.
If we later want WebSockets, we’ll likely want to just use Django
Channels.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
I added this tool a few years ago, and I did have
a vision for how it would improve our codebase, but
I can't remember exactly where I was going with it.
At this point the tool is just a little too noisy
to be helpful. An example of it creating confusion
was a recent PR where somebody was patching
user_circle_class in the PM list, and we already
had similar code in the buddy list, because they
use the same CSS. I mean, there was possibly a way
that the code could have been structured to remove
some of the duplication, but it probably would have
just moved the complexity around.
I just don't think it's worth maintaining the tool
at this point.
This experimental setting disables sending private messages in Zulip
in a crude way (i.e. users get an error when they try to send one).
It makes no effort to adjust the UI to avoid advertising the idea of
sending private messages.
Fixes#6617.
Addresses point 1 of #13533.
MissedMessageEmailAddress objects get tied to the specific that was
missed by the user. A useful benefit of that is that email message sent
to that address will handle topic changes - if the message that was
missed gets its topic changed, the email response will get posted under
the new topic, while in the old model it would get posted under the
old topic, which could potentially be confusing.
Migrating redis data to this new model is a bit tricky, so the migration
code has comments explaining some of the compromises made there, and
test_migrations.py tests handling of the various possible cases that
could arise.
This test mostly tests logic that I'm about
to remove in subsequent commits, and it's a bit
messy.
This commit removes 100% line coverage, but I
will restore that a few commits later.
We have ~5 years of proof that we'll probably never
extend Dict with more options.
Breaking the classes into makes both a little faster
(no options to check), and we remove some options
in FoldDict that are never used (from/from_array).
A possible next step is to fine-tune the Dict to use
Map internally.
Note that the TypeScript types for FoldDict are now
more specific (requiring string keys). Of course,
this isn't really enforced until we convert other
modules to TS.
Fixes this error after rebooting the host:
$ sudo ./destroy-all -f
zulip-install-bionic-41MM2
lxc-stop: zulip-install-bionic-41MM2: tools/lxc_stop.c: main: 191 zulip-install-bionic-41MM2 is not running
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
The host environment variables (especially PATH) should not be allowed
to pollute the test and could interfere with it.
This allows test-install to run on a NixOS host.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Fixes#13452.
The migration from UserProfile.is_realm_admin/UserProfile.is_guest in
e10361a832 broke our LDAP-based support
for setting a user's role via LDAP properties, which relied on setting
those fields. Because the django-auth-ldap feature powering that only
supports booleans (and in any case, we don't want to expose constants
like `ROLE_REALM_ADMINISTRATOR` to the LDAP configuration interface),
it makes sense to provide setters for these legacy fields for
backwards-compatibility.
We lint against using these setters directly in Zulip's codebase
directly. The issue with using these is that when changing user's
.role we want to create appropriate RealmAuditLog entries and send
events. This isn't possible when using these setters - the log entries
and events should be created if the role change in the UserProfile is
actually save()-ed to the database - and on the level of the setter
function, it's not known whether the change will indeed be saved.
It would have to be somehow figured out on the level of post_save
signal handlers, but it doesn't seem like a good design to have such
complexity there, for the sake of setters that generally shouldn't be
used anyway - because we prefer the do_change_is_* functions.
The purpose of this change is narrowly to handle use cases like the
setattr on these boolean properties.
A bug in Zulip's new user signup process meant that users who
registered their account using social authentication (e.g. GitHub or
Google SSO) in an organization that also allows password
authentication could have their personal API key stolen by an
unprivileged attacker, allowing nearly full access to the user's
account.
Zulip versions between 1.7.0 and 2.0.6 were affected.
This commit fixes the original bug and also contains a database
migration to fix any users with corrupt `password` fields in the
database as a result of the bug.
Out of an abundance of caution (and to protect the users of any
installations that delay applying this commit), the migration also
resets the API keys of any users where Zulip's logs cannot prove the
user's API key was not previously stolen via this bug. Resetting
those API keys will be inconvenient for users:
* Users of the Zulip mobile and terminal apps whose API keys are reset
will be logged out and need to login again.
* Users using their personal API keys for any other reason will need
to re-fetch their personal API key.
We discovered this bug internally and don't believe it was disclosed
prior to our publishing it through this commit. Because the algorithm
for determining which users might have been affected is very
conservative, many users who were never at risk will have their API
keys reset by this migration.
To avoid this on self-hosted installations that have always used
e.g. LDAP authentication, we skip resetting API keys on installations
that don't have password authentication enabled. System
administrators on installations that used to have email authentication
enabled, but no longer do, should temporarily enable EmailAuthBackend
before applying this migration.
The migration also records which users had their passwords or API keys
reset in the usual RealmAuditLog table.
This commit was automatically generated by `tools/lint --only=eslint
--fix`, after an `.eslintrc.json` change.
A half dozen files were removed from the changes by tabbott pending
further work to ensure we avoid breaking valuable PRs with merge
conflicts.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
We'll be soon documenting a production workflow that involves using
it, and that means it needs to live under scripts/ (since tools/ isn't
present in release tarballs).
Upcoming changes in test_generated_curl_examples_for_success modifies
various data of iago user heavily. So it's much easier to run
test_the_api initially than making various changes in tests of
test_the_api function.
This suppresses the mypy message “Success: no issues found in 1085
source files” or “Found 1 error in 1 file (checked 1085 source files)”
in the output of lint.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This tool doesn’t match our current workflow for Python requirements
upgrades as of commit ec9bf6576a (#13213).
It also has a type error with mypy 0.730, which would be easily fixable,
but removing it is easier.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This tool provides no value over `pip list --outdated`. It also has a
type error with mypy 0.730, which would be easily fixable, but
removing it is easier.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Then, find and fix a predictable number of previous misuses.
With a small change by tabbott to preserve backwards compatibility for
sending `yes` for the `forged` field.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
- Moves "Authentication in the development environment" from subsystems
to "development/authentication.md".
- Moves "Renumbering migrations" to a section within "Schema migrations".
Webpack code splitting will make the inclusion order of CSS files less
obvious, and we need to guarantee that these rules follow the rules
they override.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
login_context now gets the social_backends list through
get_social_backend_dicts and we move display_logo customization
to backend class definition.
This prepares for easily adding multiple IdP support in SAML
authentication - there will be a social_backend dict for each configured
IdP, also allowing display_name and icon customization per IdP.
This adds the general machinery required, and sets it up for the file
`typing_status.js` as a first use case.
Co-authored-by: Anders Kaseorg <anders@zulipchat.com>
It happens that commonmark, python-jose, and python-twitter don’t
actually use future on Python 3, and moto uses aws-xray-sdk in such a
way that it doesn’t use future, but this was a weird game to be
playing just to remove one dependency, and it caused CI failures after
new releases of future, so let’s just include it.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Previously, while Django code that relied on EXTERNAL_HOST and other
settings would know the Zulip server is actually on port 9991, the
upcoming Django SAML code in python-social-auth would end up detecting
a port of 9992 (the one the Django server is actually listening on).
We fix this using X-Forwarded-Port.
Apparently, the CircleCI and Codecov links (and the Codecov badge)
weren't pointing specifically at master, so they'd sometimes show
state from the lastest push to a pull request, which isn't a
reasonable way to advertise whether the project's build is passing.
Apparently Tornado decompresses gzip responses by default. Worse, it
fails to adjust the Content-Length header when it does.
https://github.com/tornadoweb/tornado/issues/2743
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
The reason that `pip-tools` running on Python 3 didn’t detect the
right requirements for `thumbor` on Python 2 is simply that some of
them are conditional on the Python version.
As for the requirements that had been manually added as a workaround:
`backports-abc` and `singledispatch` are now correctly detected, while
`backports.ssl-match-hostname` was vendored into `urllib3` some time
ago and `certifi` is no longer necessary.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This reverts commit 073ecaac66 (#9365).
This exception handler was overly broad in catching all `OSError`s,
and it made debugging harder by hiding the actual exception.
Furthermore, we no longer use NFS (#12963), and we’re now getting
reports of Windows users running into this message.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Otherwise Bootstrap doesn’t get minified, and also the minification
state is incorrectly reflected in the webpack cache.
The Terser plugin is used by default; we need to include it explicitly
to avoid removing it.
Switch from cssnano to clean-css because it’s noticeably faster.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
To replace DISTRIB_FAMILY, there’s now an os_families function using
the standard ID and ID_LIKE information in /etc/os-release.
Fixes#13070; fixes#13071.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
It’s about as fast as node-sass (faster, according to their
benchmarks) and more flexible. Autoprefixer is neat: we can now go
delete all our -moz-, -webkit-, etc. lines and have them autogenerated
as necessary based on .browserslistrc.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
We no longer use tsearch_extras, and the camo patch is irrelevant on
systemd systems (Xenial and newer). So we no longer need to
provide/install a PPA at all.
Closes#13027.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
Vishnu KS
YashRE42
Abhijeet Prasad Bodas
shanukun
Tim Abbott
Ganesh Pawar
Anders Kaseorg
Steve Howell
Aman Agrawal
Riken Shah
Gaurav Pandey
Alex Vandiver
m-e-l-u-h-a-n
Wesley Aptekar-Cassels
aryanshridhar
aditya-mitra
Mateusz Mandera
Sutou Kouhei
ryanreh99
Isaac Evans
Dinesh
orientor
Sumanth V Rao
Anders Kaseorg
Priyank Patel
Mohit Gupta
Gittenburg
arpit551
Aman Agrawal
Felix
Clara Dantas
Sara Gulotta
Sharif Naas
Puneeth Chaganti
sahil839
Rohitt Vashishtha
Rohan Prashant
Mateusz Mandera
Tim Abbott
Steve Howell
Abhishek-Balaji
Pranav
wowol
Greg Price
brendon
Stefan Weil
Hashir Sarwar
Chris Heald
Aditya Bansal
rht
David Rosa
Rishi Gupta
Ray Kraesig
Wyatt Hoodes