Enable numeric separators in release builds
Categories
(Core :: JavaScript Engine, enhancement, P3)
Tracking
()
People
(Reporter: till, Assigned: rhunt)
References
(Blocks 1 open bug, )
Details
(Keywords: dev-doc-complete, parity-chrome)
Attachments
(1 file)
... once we're confident there'll be no more churn in the spec.
Reporter | ||
Updated•6 years ago
|
Updated•6 years ago
|
Comment 1•6 years ago
|
||
Till, can you provide a better description into what this is supposed to be and why firefox 60 (P1) is flagged as affected?
Reporter | ||
Comment 2•6 years ago
|
||
FF60 being affected is simply the default for a new bug. It's also true in that FF60 doesn't have this enabled, but, once the blocking bug is fixed, could. That doesn't mean it should be P1 at all. As for the description, I added a link to the spec proposal as the URL, I hope that's sufficient.
Updated•6 years ago
|
Is the fix in bug 1421400 Nightly-only?
(In reply to j.j. from comment #3)
Is the fix in bug 1421400 Nightly-only?
Answering myself: Yes, seems so.
This should land in Beta 68 for Chrome compat and for 68 ESR and Fennec.
Chrome 75 ships Numeric Separators in two weeks
https://www.chromestatus.com/feature/5829906369871872
Reporter | ||
Comment 5•5 years ago
|
||
(In reply to j.j. from comment #4)
This should land in Beta 68 for Chrome compat and for 68 ESR and Fennec.
I don't think numeric separators will be used in code shipping to browsers quickly enough that we'd have to worry about Chrome compat here. Even for the ESR I really don't think that this needs to be uplifted.
What would make sense to me is to let it ride the trains either this or the next release—it seems sufficiently low-risk that we don't have to wait very long.
Updated•5 years ago
|
Assignee | ||
Comment 6•5 years ago
|
||
Spec: https://github.com/tc39/proposal-numeric-separator
This was released in V8 release 75 [1], and it sounds like spec concerns have
been figured out and we can let this ride to release.
Assignee | ||
Comment 7•5 years ago
|
||
Bug 1566516 imported tests from the WASM spec repository that use numeric separators, which caused beta simulation failures in bug 1568806. I can modify the import script to not generate '_' in numeric literals, but it sounds like this is ready to ride the trains to release.
If no one has objections, I'd like to let this ride the trains to release.
Assignee | ||
Updated•5 years ago
|
Assignee | ||
Comment 8•5 years ago
|
||
https://treeherder.mozilla.org/#/jobs?repo=try&revision=96e7ee047e71d911bb47e7e66c102fb49f287503
I'm unsure if this requires updates to test262-update.py or not. I just removed the NIGHTLY_BUILD's from the initial implementation patch.
Pushed by rhunt@eqrion.net: https://hg.mozilla.org/integration/autoland/rev/d730bc44f2d7 JS: Let numeric separators ride the trains to release. r=jorendorff
Comment 10•5 years ago
|
||
bugherder |
Comment 11•5 years ago
|
||
Doc updates:
- Added to Fx 70 release notes: https://developer.mozilla.org/en-US/docs/Mozilla/Firefox/Releases/70#JavaScript
- Added to numeric literal reference docs: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Lexical_grammar#Numeric_separators
- Added an example to
parseInt
that it won't work with numeric separators: https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/parseInt#Examples
fwiw, I think the error messages in Chrome are better.
- We mostly throw: "SyntaxError: missing digit after '_' numeric separator"
- Chrome throws:
- "Numeric separators are not allowed at the end of numeric literals" for
1_
- "Only one underscore is allowed as numeric separator" for
1__1
- "Numeric separators are not allowed at the end of numeric literals" for
- Separator after null is fine, though
- Firefox: "numeric separators '_' are not allowed in numbers that start with '0'"
- Chrome: "Numeric separator can not be used after leading 0."
:till, can you review if the docs look alright?
Also, do you think we should make our error messages better?
Reporter | ||
Comment 12•5 years ago
|
||
The docs look great! The only small nit is that "Only one underscore is allowed" is slightly ambiguous, but I don't think that's a problem in practice, as the example right below makes clear what it means.
Making the error messages better would be good, yes. Could you file a new bug for that?
Comment 14•5 years ago
|
||
Thanks for your feedback, Till!
Reworded to remove ambiguity and filed https://bugzilla.mozilla.org/show_bug.cgi?id=1589072 for potentially improving the error messages.
Updated•5 years ago
|
Updated•5 years ago
|
Description
•