feat: add aot option to karma #28854
Conversation
| }) | ||
| class Hello {} | ||
|
|
||
| expect((Hello as any).ɵcmp.template.toString()).not.toContain('jit'); |
There was a problem hiding this comment.
This is fragile and sensitive to any structural changes in JIT. It could easily cause false negatives.
What would be a better way to achieve this detection?
There was a problem hiding this comment.
I think as long as we have a second test case that tests the reverse, it's fine..? That way we can make sure that "jit" appears in one branch and doesn't appear in the other, with aot: true vs aot: false as the only difference. That seems like an okay state to be in for now.
| ...BASE_OPTIONS, | ||
| aot: true, | ||
| /** Cf. {@link ../builder-mode_spec.ts} */ | ||
| polyfills: ['zone.js', '@angular/localize/init', 'zone.js/testing'], |
There was a problem hiding this comment.
Should we factorize this in BASE_OPTIONS? or create a BASE_POLYFILLS? or an object mother: optionsMother.base().withPolyfills()?
There was a problem hiding this comment.
describeKarmaBuilder already runs the builder in both modes automatically. You shouldn't need any override but aot: true in here if everything works as expected. The builder mode spec is special because it explicitly tests that the mode can be changed by explicitly setting this. But it seems wasteful to rerun this test case twice in application builder mode (which I think is what's currently happening with this file).
yjaaidi
force-pushed
the
feat/add-aot-option-to-karma
branch
from
38ac095 to
48d0a96
Compare
|
Sorry for the lack of review so far but I just want to quickly say that I'm really excited to see this option added. |
|
No rush at all! |
| ...BASE_OPTIONS, | ||
| aot: true, | ||
| /** Cf. {@link ../builder-mode_spec.ts} */ | ||
| polyfills: ['zone.js', '@angular/localize/init', 'zone.js/testing'], |
There was a problem hiding this comment.
describeKarmaBuilder already runs the builder in both modes automatically. You shouldn't need any override but aot: true in here if everything works as expected. The builder mode spec is special because it explicitly tests that the mode can be changed by explicitly setting this. But it seems wasteful to rerun this test case twice in application builder mode (which I think is what's currently happening with this file).
| import { BASE_OPTIONS, KARMA_BUILDER_INFO, describeKarmaBuilder } from '../setup'; | ||
| import { BuilderMode } from '../../schema'; | ||
|
|
||
| describeKarmaBuilder(execute, KARMA_BUILDER_INFO, (harness, setupTarget) => { |
There was a problem hiding this comment.
If you want to restrict this to application builder mode only, you might need to add the 3rd argument for this callback (isApplicationBuilder) and skip/disable the test if that boolean is false.
| }) | ||
| class Hello {} | ||
|
|
||
| expect((Hello as any).ɵcmp.template.toString()).not.toContain('jit'); |
There was a problem hiding this comment.
I think as long as we have a second test case that tests the reverse, it's fine..? That way we can make sure that "jit" appears in one branch and doesn't appear in the other, with aot: true vs aot: false as the only difference. That seems like an okay state to be in for now.
|
Thx! No problem at all. |
PR Checklist
Please check to confirm your PR fulfills the following requirements:
PR Type
What kind of change does this PR introduce?
What is the current behavior?
Tests run in JIT only.
What is the new behavior?
Tests can run in AOT, using the new
aotoption.Does this PR introduce a breaking change?
Other information