We can relax this restriction in the future, but
basically every time it came up for me, the test
code was just disorganized, or it had an easy
workaround.
We still need to write to these globals with set_global because the
code being tested reads from them, but the tests themselves should
never need to read from them.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
It's actually pretty rare in our codebase to
call methods like `$(...).map` or `$(...).each`,
but we now support them better in zjquery.
You can pass a list of child elements now to
`$.create(...)`.
Now when we want to measure how long a block
of code takes to execute, we just wrap it with
`blueslip.measure_time`, instead of the awkward
idiom from my original commit of getting a callback
function.
My rationale for the original scheme was that I
wanted to minimize diffs and avoid changing
`const` to `let` in a few cases, but I believe
now that the function wrapper is nicer.
In a few cases I just removed the blueslip timing
code, since I was able to confirm on czo that
the times were pretty minimal.
Instead of prohibiting ‘return undefined’ (#8669), we require that a
function must return an explicit value always or never. This prevents
you from forgetting to return a value in some cases. It will also be
important for TypeScript, which distinguishes between undefined and
void.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
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.
There is good reason to do this (explanation is bit long!). With the
TypeScript migration, and the require and ES6 migrations that come
with it, we use require instead of set_global which loads the entire
module. Suppose we have a util module, which is used by some other
module, say message_store, and util is being required in message_store
since it is removed from window. Then, if a test zrequires
message_store first, and then zrequires the util module qand mocks one
of its methods, it will not be mocked for the message_store
module. The reason is:
1. zrequire('message_store') leads to require('util').
2. zrequire('util') removes the util module from cache and it is
reloaded. Now the util module in message_store and the one in
the test will be different and any updates to it in tests won't
be reflected in the actual code.
Which can lead to confusion for folks writing tests. I'll mention this
can be avoided doing zrequire('util') first but...that is not ideal.
And, since there was one outlier test that relied on this behavior,
we add the namespace.reset_module function.
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>
This just lets us temporarily assign a value
to a field.
Differences with the "override" scheme:
* override only works on globals
* override (when passed in via run_test) will
just automatically clean up at the end of
the function
We want to undo overrides in reverse order,
which is important if you override the
same name more than once in the same
function.
Until today the code basically prevented
us from ever using the original implementation
of a name we stubbed, and most of them start
as undefined due to their parent modules
starting with `set_global`.
But I do want this proper, and I introduced
a tiny pitfall today.
There was only one place where we weren't
overriding a function, and the use case there
was fairly unique.
Knowing that we're dealing with only functions
will simplify override and allow us to add
features like detecting spurious stubs.
This forces us to more explicitly document at the
top of the file what dependencies we are stubbing,
plus it's less magical.
Also, we may want to do occasional audits of
set_global to clean up places where we mock
things like stream_data, which are probably just
easier to use the real version of now that we
have cleaner APIs to set up stream data.
The modules most affected by this change are our
dispatch-oriented tests--basically, all the
modules that test handling of Zulip events
plus hotkey.js.
Before we were making it impossible to reuse
the function again (so we were preventing
leaks), but it's fine to just restore the
original function, especially now that some
of our tests have grown bigger.
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>
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>
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>
We are trying to phase out the trigger-event way
of telling modules to do something.
In this case we not only remove the indirection
of the event handler, but we also get to remove
`compose_fade` from the `ui_init` startup sequence.
This also has us update `compose_fade` outside
the loop, although that's only a theoretical
improvement, since I don't think `peer_add` events
every actually include multiple streams.
To make the dispatch tests a little flatter, I
added a one-line change to zjsunit to add
`make_stub` to `global`.
To manually test:
* have Aaron reply to Denmark (keep compose box open)
* have Iago add Hamlet to Denmark
* have Hamlet unsubscribe
* Add action to mute topics.
* We don't need to store muted data per topic as previously planned.
* Moved launch topic test to the top so that they run on non-modified
data.
JSON.parse behaves as we want for numbers but for strings, we would
throw an error like 'unexpected token at position 0'. This meant we
couldn't read back the value set by `$input.data('val', 'text')`.
We change the user facing interface to allow specifying expected
number of error messages (default=1). Now an average test can look
like:
```
// We expect 3 error messages;
blueslip.expect('error', 'an error message', 3);
throwError();
throwError();
throwError();
blueslip.reset();
```
We now use `assert.throws()` to test that we're
properly calling `blueslip.fatal`.
In order to not break line coverage here, we have
to remove an unreachable `return` in `stream_data.js`.
Usually we test `fatal` for line coverage reasons.
Most places where we use `blueslip.fatal` fall in
these categories:
* the code is theoretically unreachable, but
we have `blueslip.fatal` for defensive reasons
* we have some upstream bug that we should just
fix
* the code should recover gracefully and just
use blueslip.errors()
It's possible that we should eliminate `blueslip.fatal`
from our API and just throw errors when really important
invariants get broken. This will make it more obvious
to somebody reading the code that we're not going to
continue after the call, and `blueslip` already knows
how to catch exceptions and report them.
Explicitly stubbing i18n in 48 different files
is mostly busy work at this point, and it doesn't
provide much signal, since often it's invoked
only to satisfy transitive dependencies.
Let's say you have module hello.js like so:
// hello.js
const hello_world = i18n.t('Hello world');
exports.get_greeting = () => hello_world;
And then two modules like this:
// apple.js
const hello = require('hello');
exports.foo = () => {
show_greeting(hello.get_greeting());
};
// banana.js
const hello = require('hello');
exports.foo = () => {
display_greeting(hello.get_greeting());
};
The test for apple.js could look like this,
and it won't crash due to the stub:
set_global('i18n', {t: () => {}});
zrequire('hello');
zrequire('apple');
Now let's say your write this broken version
of a test for banana.js:
zrequire('hello');
zrequire('banana');
If you run `./tools/test-js-with-node`, the
"banana" test will pass, because while it
does require "hello", it won't actually
*execute* the code that happens at require
time for "hello", because it's already in
the cache. Here is the code that gets
skipped:
const hello_world = i18n.t('Hello world');
But then if you try to run the banana test
individually, the above line of code will
cause the test to crash. And it will crash
even before you actually try to test the
meaningful code here:
exports.foo = () => {
display_greeting(hello.get_greeting());
};
This commit fixes this leak scenario by just
aggressively clearing out things from the
require cache.
This slows tests down by about 10%, which I think
is worth the extra safety here.
This is not always a behavior-preserving translation: _.extend mutates
its first argument. However, the code does not always appear to have
been written to expect that.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
This is not always a behavior-preserving translation: $.extend mutates
its first argument. However, the code does not always appear to have
been written to expect that.
Signed-off-by: Anders Kaseorg <anders@zulipchat.com>
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>
Steve Howell
Anders Kaseorg
Priyank Patel
Aman Agrawal
Rohitt Vashishtha
Mohit Gupta
Sara Gulotta
Steve Howell
Stefan Weil
Anders Kaseorg