Uh oh!
There was an error while loading. Please reload this page.
Enforce a size limit in getSpreadType - #40755
Conversation
When a union is spread into a union, the sizes are multiplied, potentially resulting in an enormous union (especially if there are repeated spreads). This check detects cases that used to run out of memory. Fixesmicrosoft#40754
Andrew Casey (amcasey)
commented
Sep 24, 2020
TypeScript Bot (@typescript-bot) user test this |
Heya Andrew Casey (@amcasey), I've started to run the parallelized community code test suite on this PR at 6650496. You can monitor the build here. |
| const resultSize = (left as UnionType).types.length * (right as UnionType).types.length; | ||
| if (resultSize > 100000) { | ||
| tracing.instant(tracing.Phase.Check, "getSpreadType_DepthLimit", { leftId: left.id, rightId: right.id }); | ||
| error(currentNode, Diagnostics.Spread_expression_produces_a_union_type_that_is_too_complex_to_represent); |
There was a problem hiding this comment.
I think this error is going to be frustrating to encounter with it also providing a suggestion for a workaround. What is the general-form workaround? Casting something somewhere to any? Should we add a related span/quickfix for that?
There was a problem hiding this comment.
I don't think there's a fix I'm sufficiently confident in to make it a quick fix. I gave this a new error number so that it would be easier to document/google.
There was a problem hiding this comment.
I did wonder about looking inside the union to figure out whether there's a smaller span we could squiggle.
There was a problem hiding this comment.
For the real code where this arose, the mitigation I suggested was adding a cast to each spread with a pre-existing type that listed all possible properties (optionally). I imagine any would also work, possible at the expense of completions and such.
Wesley Wigham (weswigham)
left a comment
There was a problem hiding this comment.
This code is definitely fine, and, even if we broaden #34853 to fix the reported issue without an error (which I think we should consider), we'll still fundamentally need this as a fallback.
Andrew Casey (amcasey)
commented
Sep 24, 2020
If we broaden #34853, I'd probably drop the limit substantially. I based it on what would still compile quickly in my toy example. |
Andrew Casey (amcasey)
commented
Sep 24, 2020
TypeScript Bot (@typescript-bot) test this |
Heya Andrew Casey (@amcasey), I've started to run the extended test suite on this PR at 6650496. You can monitor the build here. |
TypeScript Bot (typescript-bot)
commented
Sep 24, 2020
The user suite test run you requested has finished and failed. I've opened a PR with the baseline diff from master. |
Andrew Casey (amcasey)
commented
Sep 24, 2020
Since there are no occurrences of my new error code in the user test baseline diff, I'm assuming I can ignore the failure. |
TypeScript Bot (@typescript-bot) pack this |
Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the tarball bundle task on this PR at 6650496. You can monitor the build here. |
Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the perf test suite on this PR at 6650496. You can monitor the build here. Update: The results are in! |
TypeScript Bot (@typescript-bot) test this |
Heya Daniel Rosenwasser (@DanielRosenwasser), I've started to run the extended test suite on this PR at 6650496. You can monitor the build here. |
| if (resultSize > 100000) { | ||
| tracing.instant(tracing.Phase.Check, "getSpreadType_DepthLimit", { leftId: left.id, rightId: right.id }); | ||
| error(currentNode, Diagnostics.Spread_expression_produces_a_union_type_that_is_too_complex_to_represent); | ||
| return errorType; |
There was a problem hiding this comment.
Mine Starks (@minestarks) raised an interesting point: should this be anyType so that users have the option of ts-ignoring this error?
Hey Daniel Rosenwasser (@DanielRosenwasser), I've packed this into an installable tgz. You can install it for testing by referencing it in your and then running There is also a playground for this build. |
TypeScript Bot (typescript-bot)
commented
Sep 25, 2020
Daniel Rosenwasser (@DanielRosenwasser) Here they are:Comparison Report - master..40755
System
Hosts
Scenarios
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Andrew Casey (amcasey)
commented
Sep 25, 2020
This should probably use |
When a union is spread into a union, the sizes are multiplied,
potentially resulting in an enormous union (especially if there are
repeated spreads). This check detects cases that used to run out of
memory.
Fixes#40754