This PR fixes a critical performance issue where the language
service makes a MASSIVE number of filesystem calls when performing
module resolution.
This is because there is no caching. To make matters worse, module
resolution is performed for **every** program change (which means every
few keystrokes trigger a massive number of fs calls).
PR Close#32479
Prior to this commit, a directive with a selector `selector=".Titledir"` would not match an element like `div class="titleDir"` in Ivy whereas it would in VE. The same issue was present for `selector="[title=Titledir]` and `title="titleDir"`. This fixes the Ivy behavior by changing the matching algorithm to use lowercased values.
Note that some `render3` tests needed to be changed to reflect that the compiler generates lowercase selectors. These tests are in the process to be migrated to `acceptance` to use `TestBed` in another PR anyway.
PR Close#32548
Adds two acceptance tests to show a current difference in behavior between Ivy and VE.
A directive with a selector `.Titledir` matches an element with `class="titleDir"` in VE but not in Ivy.
Same thing for an attribute value.
PR Close#32548
Prior to this commit, the `previousOrParentTNode` was set to null after performing all operations within `refreshView` function. It's causing problems in more complex scenarios, for example when change detection is triggered during DI (see test added as a part of this commit). As a result, global state might be corrupted. This commit captures current value of `previousOrParentTNode` and restores it after `refreshView` call.
PR Close#32521
Replaces the `select` instruction with a new one called `advance`. Instead of the jumping to a specific index, the new instruction goes forward X amount of elements. The advantage of doing this is that it should generate code the compresses better.
PR Close#32516
The `goog.getMsg()` function requires placeholder names to be camelCased.
This is not the case for `$localize`. Here placeholder names need
match what is serialized to translation files.
Specifically such placeholder names keep their casing but have all characters
that are not in `a-z`, `A-Z`, `0-9` and `_` converted to `_`.
PR Close#32509
Prior to this change, the template source mapping details were always
built during the analysis phase, under the assumption that pre-analysed
templates would always correspond with external templates. This has
turned out to be a false assumption, as inline templates are also
pre-analyzed to be able to preload any stylesheets included in the
template.
This commit fixes the bug by capturing the template source mapping
details at the moment the template is parsed, which is either during the
preanalysis phase when preloading is available, or during the analysis
phase when preloading is not supported.
Tests have been added to exercise the template error mapping in
asynchronous compilations where preloading is enabled, similar to how
the CLI performs compilations.
Fixes#32538
PR Close#32544
Adds support for `styleUrls` definitions in the same way `templateUrl`
definitions are provided; clicking on styleUrl will take a user to the
respective file.
Unifies some code in determining a URL definition. We first check if a
url is a `templateUrl`; if it's not, we check that it's a `styleUrl` or
return no definitions.
PR Close#32464
This gives an overview of how much time is spent in each operation/phase
and makes it easy to do rough comparisons of how different
configurations or changes affect performance.
PR Close#32427
`ngcc` supports both synchronous and asynchronous execution. The default
mode when using `ngcc` programmatically (which is how `@angular/cli` is
using it) is synchronous. When running `ngcc` from the command line
(i.e. via the `ivy-ngcc` script), it runs in async mode.
Previously, the work would be executed in the same way in both modes.
This commit improves the performance of `ngcc` in async mode by
processing tasks in parallel on multiple processes. It uses the Node.js
built-in [`cluster` module](https://nodejs.org/api/cluster.html) to
launch a cluster of Node.js processes and take advantage of multi-core
systems.
Preliminary comparisons indicate a 1.8x to 2.6x speed improvement when
processing the angular.io app (apparently depending on the OS, number of
available cores, system load, etc.). Further investigation is needed to
better understand these numbers and identify potential areas of
improvement.
Inspired by/Based on @alxhub's prototype: alxhub/angular@cb631bdb1
Original design doc: https://hackmd.io/uYG9CJrFQZ-6FtKqpnYJAA?view
Jira issue: [FW-1460](https://angular-team.atlassian.net/browse/FW-1460)
PR Close#32427
This commit adds a new `TaskQueue` implementation that supports
executing multiple tasks in parallel (while respecting interdependencies
between them).
This new implementation is currently not used, thus the behavior of
`ngcc` is not affected by this change. The parallel `TaskQueue` will be
used in a subsequent commit that will introduce parallel task execution.
PR Close#32427
This change does not alter the current behavior, but makes it easier to
introduce `TaskQueue`s implementing different task selection algorithms,
for example to support executing multiple tasks in parallel (while
respecting interdependencies between them).
Inspired by/Based on @alxhub's prototype: alxhub/angular@cb631bdb1
PR Close#32427
Previously, `ngcc` needed to store some metadata related to the
processing of each entry-point. This metadata was stored in a `Map`, in
the form of `EntryPointProcessingMetadata` and passed around as needed.
After some recent refactorings, it turns out that this metadata (with
its only remaining property, `hasProcessedTypings`) was no longer used,
because the relevant information was extracted from other sources (such
as the `processDts` property on `Task`s).
This commit cleans up the code by removing the unused code and types.
PR Close#32427
In the past, a task's processability didn't use to be known in advance.
It was possible that a task would be created and added to the queue
during the analysis phase and then later (during the compilation phase)
it would be found out that the task (i.e. the associated format
property) was not processable.
As a result, certain checks had to be delayed, until a task's processing
had started or even until all tasks had been processed. Examples of
checks that had to be delayed are:
- Whether a task can be skipped due to `compileAllFormats: false`.
- Whether there were entry-points for which no format at all was
successfully processed.
It turns out that (as made clear by the refactoring in 9537b2ff8), once
a task starts being processed it is expected to either complete
successfully (with the associated format being processed) or throw an
error (in which case the process will exit). In other words, a task's
processability is known in advance.
This commit takes advantage of this fact by moving certain checks
earlier in the process (e.g. in the analysis phase instead of the
compilation phase), which in turn allows avoiding some unnecessary work.
More specifically:
- When `compileAllFormats` is `false`, tasks are created _only_ for the
first suitable format property for each entry-point, since the rest of
the tasks would have been skipped during the compilation phase anyway.
This has the following advantages:
1. It avoids the slight overhead of generating extraneous tasks and
then starting to process them (before realizing they should be
skipped).
2. In a potential future parallel execution mode, unnecessary tasks
might start being processed at the same time as the first (useful)
task, even if their output would be later discarded, wasting
resources. Alternatively, extra logic would have to be added to
prevent this from happening. The change in this commit avoids these
issues.
- When an entry-point is not processable, an error will be thrown
upfront without having to wait for other tasks to be processed before
failing.
PR Close#32427
Previously, `ngcc`'s programmatic API would run and complete
synchronously. This was necessary for specific usecases (such as how the
`@angular/cli` invokes `ngcc` as part of the TypeScript module
resolution process), but not for others (e.g. running `ivy-ngcc` as a
`postinstall` script).
This commit adds a new option (`async`) that enables turning on
asynchronous execution. I.e. it signals that the caller is OK with the
function call to complete asynchronously, which allows `ngcc` to
potentially run in a more efficient mode.
Currently, there is no difference in the way tasks are executed in sync
vs async mode, but this change sets the ground for adding new execution
options (that require asynchronous operation), such as processing tasks
in parallel on multiple processes.
NOTE:
When using the programmatic API, the default value for `async` is
`false`, thus retaining backwards compatibility.
When running `ngcc` from the command line (i.e. via the `ivy-ngcc`
script), it runs in async mode (to be able to take advantage of future
optimizations), but that is transparent to the caller.
PR Close#32427
This change does not alter the current behavior, but makes it easier to
introduce new types of `Executors` , for example to do the required work
in parallel (on multiple processes).
Inspired by/Based on @alxhub's prototype: alxhub/angular@cb631bdb1
PR Close#32427
To persist some of its state, `ngcc` needs to update `package.json`
files (both in memory and on disk).
This refactoring abstracts these operations behind the
`PackageJsonUpdater` interface, making it easier to orchestrate them
from different contexts (e.g. when running tasks in parallel on multiple
processes).
Inspired by/Based on @alxhub's prototype: alxhub/angular@cb631bdb1
PR Close#32427
In order to prevent `ngcc`'d packages (e.g. libraries) from getting
accidentally published, `ngcc` overwrites the `prepublishOnly` npm
script to log a warning and exit with an error. In case we want to
restore the original script (e.g. "undo" `ngcc` processing), we keep a
backup of the original `prepublishOnly` script.
Previously, running `ngcc` a second time (e.g. for a different format)
would create a backup of the overwritten `prepublishOnly` script (if
there was originally no `prepublishOnly` script). As a result, if we
ever tried to "undo" `ngcc` processing and restore the original
`prepublishOnly` script, the error-throwing script would be restored
instead.
This commit fixes it by ensuring that we only back up a `prepublishOnly`
script, iff it is not the one we created ourselves (i.e. the
error-throwing one).
PR Close#32427
This PR partially fixes a circular dependency problem whereby the
creation of a project queries Angular plugin for external files, but
the discovery of external files requires root files to be defined in a
Project. The right approach is to return empty array if Project has no
root files.
PR Close#32519
The commit range that is associated with a CI build is used for a couple
of things (mostly related to payload-size tracking):
- Determine whether a size change was caused by application code or
dependencies (or both).
- Add the messages of the commits associated with the build (and thus
the payload-size change).
NOTE: The commit range is only used on push builds.
Previously, the commit range was computed based on the
`CIRCLE_COMPARE_URL` environment variable. With [CircleCI Pipelines][1]
enabled, `CIRCLE_COMPARE_URL` is no longer available and the commit
range cannot be reliably detected.
This commit switches `CI_COMMIT_RANGE` to only include the last commit.
This can be less accurate in some rare cases, but is true in the
majority of cases (on push builds). Additionally, it stores the CircleCI
build URL in the database along with the payload data, so the relevant
info can be retrieved when needed.
[1]: https://circleci.com/docs/2.0/build-processing
PR Close#32537
The Language Service integration tests should reinstall the Angular
distribution every time it is built. Adds a `yarn build-dist`
convinience script so building the distribution doesn't have to
happen on the repo root. This new script also refreshes the installed
modules. Building is expesnive, so it is not bundled with testing
scripts.
PR Close#32515
Logs a warning instead of throwing when running into a binding to an unknown property in JIT mode. Since we aren't using a schema for the runtime validation anymore, this allows us to support browsers where properties are unsupported.
PR Close#32463
The docs examples e2e tests have been quite flaky recently. This causes
the `test_docs_examples` and `test_docs_examples_ivy` CircleCI jobs to
often fail (and block PRs) without a real reason.
This commit adds support for re-running failed docs examples e2e tests
and configures the `test_docs_examples` and `test_docs_examples_ivy`
jobs to try running each test that fails a second time, before giving up
and marking it as failed.
Closes#31841Closes#31842
PR Close#32497
Prior to this commit, listeners order was not preserved in case we coalesce them to avoid triggering unnecessary change detection cycles. For performance reasons we were attaching listeners to existing events at head (always using first listener as an anchor), to avoid going through the list, thus breaking the order in which listeners are registered. In some scenarios this order might be important (for example with `ngModelChange` and `input` listeners), so this commit updates the logic to put new listeners at the end of the list. In order to avoid performance implications, we keep a pointer to the last listener in the list, so adding a new listener takes constant amount of time.
PR Close#32484
We need to be clearer to developers who upgrade to v9 (next) and get this
error, why they have a problem and what they have to do about it.
Once we have a better CLI schematics story, where this import will be
included by default in new applications and a CLI migration will add it
when upgrading apps to v9, we could simplify or remove this error message.
PR Close#32491
Fixes an issue where Ivy incorrectly inserts items in the beginning of an `ngFor`, if the `ngFor` is set on an `ng-container`. The issue comes from the fact that we choose the `ng-container` comment node as the anchor point before which to insert the content, however the node might be after any of the nodes inside the container. These changes switch to picking out the first node inside of the container instead.
PR Close#32324
Historically bind() used to be a separate instruction. With a series of
refactoring it became a utility function but after recent code changes
it does not provide any valuable abstraction / help. On the contrary -
it can be seen as a performance problem (forces unnecessary comparison to
`NO_CHANGE` during change detection).
PR Close#32489
Andrew Scott
Pawel Kozlowski
Kapunahele Wong
Pete Bacon Darwin
Tom Kwong
cexbrayat
Trevor Karjanis
George Kalpakas
Keen Yee Liau
Andrew Kushnir
crisbeto
Phaneendra
Carlos Ortiz García
JoostK
ayazhafiz
Leonardo Zizzamia
lazarljubenovic
Paul Gschwendtner
Keen Yee Liau
Greg Magolan
Dominik Pieper
Timar
Joey Perrott