Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Limit unsound indexed access type relations #27490

Merged
merged 4 commits into from
Oct 3, 2018
Merged

Conversation

ahejlsberg
Copy link
Member

Fixes #27470.

Copy link
Member

@weswigham weswigham left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

While I wholeheartedly agree with chipping away at how unsound we are here (and this looks fine in that regard), I do worry about the knock-on effects in our ecosystem (especially around, eg, react). What I've seen is that when types start getting complex (eg, generic indexes + other generic type constructors) people stop understanding them and simply (over)specify them until they work without quite understanding what shape/procedure they've described. We should probably do a diff on DT with this change to see if it's as bad as I think it might be, or if I'm overthinking it. Not to see if it's a good change, but to see if we need to identify and fix any libraries which embed a similar problem to this and accidentally or intentionally rely on it before we ship it in a full release.

x1 = x2;
x1 = x3;
x1 = x4;
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Might be worth keeping all these variants (maybe in a different file or function) just to document that they're expected to fail now (and so we know if they suddenly start succeeding again).

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

They're basically covered by the new tests in keyofAndIndexedAccessErrors.ts, so I think it is fine to get rid of them.

@ahejlsberg
Copy link
Member Author

@weswigham Do we have an automated way of running the DT tests? (Similar to the RWC bot.) Otherwise, what's the best way to go about it?

@weswigham
Copy link
Member

@ahejlsberg No, the bot doesn't have a build for baselining dt. I can probably whip one up, though I don't think it'll be able to have an automatic PR/diff viewing system like RWC does. In the absence of that, running jake runtests-parallel ru=dt on master, accepting the initial baseline snapshot, then running it again on this branch would be a way to go about it locally; but you probably want someone with a real beefy machine to do it, since they take quite a long time compared to RWC, even.

@ahejlsberg
Copy link
Member Author

but you probably want someone with a real beefy machine to do it

Like yours? (Hint, hint)

@weswigham
Copy link
Member

I knew what I was getting into when I said what I said. XD

@weswigham
Copy link
Member

weswigham commented Oct 2, 2018

And 4 hours later the results are in. My fears seem overblown, this change doesn't seem to affect anything on DT. For reference, my motivating issue for my worry was #27201, wherein what people are trying to do might be affected by this.

@typescript-bot test this just in case tho

@typescript-bot
Copy link
Collaborator

typescript-bot commented Oct 2, 2018

Heya @weswigham, I've started to run the extended test suite on this PR at 69cd6c0. You can monitor the build here. It should now contribute to this PR's status checks.

@squirly
Copy link

squirly commented Oct 3, 2018

I just tried this in my project, and it indeed does fix #27470! 🎉

@weswigham
Copy link
Member

I think rwc on this PR is fine; this PR is just missing a change in master (semicolon in a helper) which caused an RWC change. Everything looks good.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

5 participants