Root cause is that for perf reasons we cache `LFrame` so that we don't have to allocate it all the time. To be extra fast we clear the `LFrame` on `enterView()` rather that on `leaveView()`. The implication of this strategy is that the deepest `LFrame` will retain objects until the `LFrame` allocation depth matches the deepest object.
The fix is to simply clear the `LFrame` on `leaveView()` rather then on `enterView()`
Fix#35148
PR Close#35156
There are different `DebugNode`/`DebugElement` implementations (and
associated helper functions) for ViewEngine and Ivy. Additionally, these
classes/functions, which are defined inside the `core` package, are
imported by the `platform-browser` package.
Previously, this code was not tree-shaken as expected in Ivy. #30130
partially addressed the issue, but only for the case where `core` and
`platform-browser` end up in the same closure after webpack's scope
hoisting. In cases where this is not the case, our webpack/terser based
tooling is not capable of tree-shaking it.
This commit fixes the problem, by ensuring that the code retained in Ivy
mode (due to the cross-package import) does not unnecessarily reference
`DebugNode`/`DebugElement`, allowing the code to be tree-shaken away.
This results in a 7.6KB reduction in the size of the main angular.io
bundle.
Jira issue: [FW-1802](https://angular-team.atlassian.net/browse/FW-1802)
PR Close#35003
Previously we would write to class/style as strings `element.className` and `element.style.cssText`. Turns out that approach is good for initial render but not good for updates. Updates using this approach are problematic because we have to check to see if there was an out of bound write to style and than perform reconciliation. This also requires the browser to bring up CSS parser which is expensive.
Another problem with old approach is that we had to queue the DOM writes and flush them twice. Once on element advance instruction and once in `hostBindings`. The double flushing is expensive but it also means that a directive can observe that styles are not yet written (they are written after directive executes.)
The new approach uses `element.classList.add/remove` and `element.style.setProperty/removeProperty` API for updates only (it continues to use `element.className` and `element.style.cssText` for initial render as it is cheaper.) The other change is that the styling changes are applied immediately (no queueing). This means that it is the instruction which computes priority. In some circumstances it may result in intermediate writes which are than overwritten with new value. (This should be rare)
Overall this change deletes most of the previous code and replaces it with new simplified implement. The simplification results in code savings.
PR Close#34804
NOTE: This change must be reverted with previous deletes so that it code remains in build-able state.
This change deletes old styling code and replaces it with a simplified styling algorithm.
The mental model for the new algorithm is:
- Create a linked list of styling bindings in the order of priority. All styling bindings ere executed in compiled order and than a linked list of bindings is created in priority order.
- Flush the style bindings at the end of `advance()` instruction. This implies that there are two flush events. One at the end of template `advance` instruction in the template. Second one at the end of `hostBindings` `advance` instruction when processing host bindings (if any).
- Each binding instructions effectively updates the string to represent the string at that location. Because most of the bindings are additive, this is a cheap strategy in most cases. In rare cases the strategy requires removing tokens from the styling up to this point. (We expect that to be rare case)S Because, the bindings are presorted in the order of priority, it is safe to resume the processing of the concatenated string from the last change binding.
PR Close#34616
The `computeStaticStyling` will be used for computing static styling value during `firstCreatePass`.
The function takes into account static styling from the template as well as from the host bindings. The host bindings need to be merged in front of the template so that they have the correct priority.
PR Closes#34418
This change moves information from instructions to declarative position:
- `ɵɵallocHostVars(vars)` => `DirectiveDef.hostVars`
- `ɵɵelementHostAttrs(attrs)` => `DirectiveDef.hostAttrs`
When merging directives it is necessary to know about `hostVars` and `hostAttrs`. Before this change the information was stored in the `hostBindings` function. This was problematic, because in order to get to the information the `hostBindings` would have to be executed. In order for `hostBindings` to be executed the directives would have to be instantiated. This means that the directive instantiation would happen before we had knowledge about the `hostAttrs` and as a result the directive could observe in the constructor that not all of the `hostAttrs` have been applied. This further complicates the runtime as we have to apply `hostAttrs` in parts over many invocations.
`ɵɵallocHostVars` was unnecessarily complicated because it would have to update the `LView` (and Blueprint) while existing directives are already executing. By moving it out of `hostBindings` function we can access it statically and we can create correct `LView` (and Blueprint) in a single pass.
This change only changes how the instructions are generated, but does not change the runtime much. (We cheat by emulating the old behavior by calling `ɵɵallocHostVars` and `ɵɵelementHostAttrs`) Subsequent change will refactor the runtime to take advantage of the static information.
PR Close#34683
Fixes Ivy detecting changes inside child embedded views, even though they're detached.
Note that there's on subtlety here: I made the changes inside `refreshDynamicEmbeddedViews` rather than `refreshView`, because we support detecting changes on a detached view (evidenced by a couple of unit tests), but only if it's triggered directly from the view's `ChangeDetectorRef`, however we shouldn't be detecting changes in the detached child view when something happens in the parent.
Fixes#34816.
PR Close#34846
Since I was learning the codebase and had a hard time understanding what was going on I've done a
bunch of changes in one commit that under normal circumstances should have been split into several
commits. Because this code is likely going to be overwritten with Misko's changes I'm not going to
spend the time with trying to split this up.
Overall I've done the following:
- I processed review feedback from #34307
- I did a bunch of renaming to make the code easier to understand
- I refactored some internal functions that were either inefficient or hard to read
- I also updated lots of type signatures to correct them and to remove many casts in the code
PR Close#34307
We should only generate the `providedIn` property in injectable
defs if it has a non-null value. `null` does not communicate
any information to the runtime that isn't communicated already
by the absence of the property.
This should give us some modest code size savings.
PR Close#34116
For injectables, we currently generate a factory function in the
injectable def (prov) that delegates to the factory function in
the factory def (fac). It looks something like this:
```
factory: function(t) { return Svc.fac(t); }
```
The extra wrapper function is unnecessary since the args for
the factory functions are the same. This commit changes the
compiler to generate this instead:
```
factory: Svc.fac
```
Because we are generating less code for each injectable, we
should see some modest code size savings. AIO's main bundle
is about 1 KB smaller.
PR Close#34076
Previously, the Angular AOT compiler would always add a
`ɵprov` to injectables. But in ngcc this resulted in duplicate `ɵprov`
properties since published libraries already have this property.
Now in ngtsc, trying to add a duplicate `ɵprov` property is an error,
while in ngcc the additional property is silently not added.
// FW-1750
PR Close#34085
This commit fixes a compatibility bug where pre-order lifecycle
hooks (onInit, doCheck, OnChanges) for directives on the same
host node were executed based on the order the directives were
matched, rather than the order the directives were instantiated
(i.e. injection order).
This discrepancy can cause issues with forms, where it is common
to inject NgControl and try to extract its control property in
ngOnInit. As the NgControl directive is injected, it should be
instantiated before the control value accessor directive (and
thus its hooks should run first). This ensures that the NgControl
ngOnInit can set up the form control before the ngOnInit
for the control value accessor tries to access it.
Closes#32522
PR Close#34026
The size diff threshold of 1% has proven to be too lenient for us
to catch size regressions in AIO. Since the AIO main bundle is
between 400-500 KB, a size regression must be between 4-5 KB before
it will cause the tests to fail. As a result, we may merge many
changes with smaller regressions of a few KB before the size test
eventually lets us know that the number has increased. The hope is
that lowering the threshold will help us catch the smaller
regressions during code review and prevent the size tests failing at
a random later time when someone catches the size "hot potato".
PR Close#33969
This commit transforms the setClassMetadata calls generated by ngtsc from:
```typescript
/*@__PURE__*/ setClassMetadata(...);
```
to:
```typescript
/*@__PURE__*/ (function() {
setClassMetadata(...);
})();
```
Without the IIFE, terser won't remove these function calls because the
function calls have arguments that themselves are function calls or other
impure expressions. In order to make the whole block be DCE-ed by terser,
we wrap it into IIFE and mark the IIFE as pure.
It should be noted that this change doesn't have any impact on CLI* with
build-optimizer, which removes the whole setClassMetadata block within
the webpack loader, so terser or webpack itself don't get to see it at
all. This is done to prevent cross-chunk retention issues caused by
webpack's internal module registry.
* actually we do expect a short-term size regression while
https://github.com/angular/angular-cli/pull/16228
is merged and released in the next rc of the CLI. But long term this
change does nothing to CLI + build-optimizer configuration and is done
primarly to correct the seemingly correct but non-function PURE annotation
that builds not using build-optimizer could rely on.
PR Close#33337
This patch gets rid of the configuration settings present in the
`TStylingContext` array that is used within the styling algorithm
for `[style]`, `[style.prop]`, `[class]` and `[class.name]` bindings.
These configurations now all live inside of the `TNodeFlags`.
PR Close#33540
Several commits merged into master recently resulted in minor payload size improvement. This commit updates the payload size limit to make corresponding CI checks pass.
PR Close#33587
This patch introduces a `firstUpdatePass` flag which can be used inside
of instruction code to determine if this is the first time each
instruction is running inside of the update block of a template or
a hostBindings function.
PR Close#31270
`bindingIndex` stores the current location of the bindings in the
template function. Because it used to be stored in `LView` that `LView`
was not reentrant. This could happen if a binding was a getter and had
a side-effect of calling `detectChanges()`.
By moving the `bindingIndex` to `LFrame` where all of the global state
is kept in reentrant way we correct the issue.
PR Close#33235
`LFrame` stores information specifice to the current `LView` As the code
enters and leaves `LView`s we use `enterView()` and `leaveView()`
respectively to build a a stack of `LFrame`s. This allows us to easily
restore the previous `LView` instruction state.
PR Close#33178
Turns out that writing to global state is more expensive than writing to
a property on an object.
Slower:
````
let count = 0;
function increment() {
count++;
}
```
Faster:
````
const state = {
count: 0
};
function increment() {
state.count++;
}
```
This change moves all of the instruction state into a single state object.
`noop_change_detection` benchmark
Pre refactoring: 16.7 us
Post refactoring: 14.523 us (-13.3%)
PR Close#33093
Removes the `Renderer` and related symbols which have been deprecated since version 4.
BREAKING CHANGES:
* `Renderer` has been removed. Use `Renderer2` instead.
* `RenderComponentType` has been removed. Use `RendererType2` instead.
* `RootRenderer` has been removed. Use `RendererFactory2` instead.
PR Close#33019
Prior to this patch, each time `advance()` would run (or when a
templateFn or hostBindings code exits) then the core change detection
code would check to see whether the styling data needs to be reset. This
patch removes that functionality and places everything inside of the
scheduled styling exit function. This means that each time one or more
styling bindings run (even if the value hasn't changed) then an exit
function will be scheduled and that will do all the cleanup.
PR Close#32591
This patch is a final major refactor in styling Angular.
This PR includes three main fixes:
All temporary state taht is persisted between template style/class application
and style/class application in host bindings is now removed.
Removes the styling() and stylingApply() instructions.
Introduces a "direct apply" mode that is used apply prop-based
style/class in the event that there are no map-based bindings as
well as property collisions.
PR Close#32259
PR Close#32591
This patch is a final major refactor in styling Angular.
This PR includes three main fixes:
All temporary state taht is persisted between template style/class application
and style/class application in host bindings is now removed.
Removes the styling() and stylingApply() instructions.
Introduces a "direct apply" mode that is used apply prop-based
style/class in the event that there are no map-based bindings as
well as property collisions.
PR Close#32259
PR Close#32596
This patch is a final major refactor in styling Angular.
This PR includes three main fixes:
All temporary state taht is persisted between template style/class application
and style/class application in host bindings is now removed.
Removes the styling() and stylingApply() instructions.
Introduces a "direct apply" mode that is used apply prop-based
style/class in the event that there are no map-based bindings as
well as property collisions.
PR Close#32259
Prior to this fix if a `NO_CHANGE` value was assigned to a binding, or
an interpolation value rendererd a `NO_CHANGE` value, then the presence
of that value would cause the internal counter index values to not
increment properly. This patch ensures that this doesn't happen and
that the counter/bitmask values update accordingly.
PR Close#32143
Misko Hevery
Igor Minar
George Kalpakas
Kristiyan Kostadinov
Pawel Kozlowski
Kara Erickson
Pete Bacon Darwin
Andrew Kushnir
Filipe Silva
Alex Rickabaugh
JiaLiPassion
Matias Niemelä
atscott