Flatten immediately nested conditional types in the false position

[ahejlsberg]

With this PR we flatten immediately nested conditional types in the false position, effectively treating types of the form

type Foo<...> =
    A extends B ? X :
    C extends D ? Y :
    E extends F ? Z :
    ...;

as a single construct for purposes of resolution. This should meaningfully improve the cost of resolving such types, plus it means they aren't subject to the instatiation depth limiter (they basically count as a single level).

Fixes #28663.

[ajafff]

I'm pretty sure we have bug for this somewhere

#28663

[weswigham]

I've thrown up #36584 as a much simpler alternative - there's no hard syntactic requirements (so it will take effect much more often than this change), it doesn't contort conditional instantiation with an inner loop, and allows the same usecases. In the cases where we would flatten here, we already don't create anything intermediate that isn't used to determine simplifyability (at least once distributivity is handled/bailed appropriately); so beyond acting as a trampoline to reduce stack depth in certain scenarios, I don't think the extra structure here serves much purpose, other than to greatly complicate the code. Ergo, I think the onus is really on this change to demonstrate a case where the extra structure actually saves us from a pathological performance cliff.

[ahejlsberg]

@weswigham Beyond fixing the depth limiter issue on large conditional types, the PR makes resolution faster by virtue of its shortened code path and reduces memory consumption by not storing resolved types in the caches associated with the each nested type's ConditionalRoot. The effects are not dramatic, but every little bit counts. I disagree that the loop greatly complicates the code. I think it is perfectly meaningful to think of a series of immediately nested conditionals as a single construct, and the loop reflects that.

[weswigham]

I think that unless this pattern is super-widespread (and it's not because we made it an error), it's really not worth giving this treatment to. I'd get giving this treatment to, like, nested object instantiation somehow, but syntactically nested conditionals are few and far between. There's a grand total of one type in the react .d.ts which would save one callstack depth for this. I feel a change like this change is just way too narrow to justify, at least without very concrete cases that it fixes and the effect it has on them.

[sandersn]

I don't think this change is that complicated. Disregarding whitespace changes and new comments, it's the same length as #36584 -- about 10 lines.
Both solutions seem a bit ad-hoc to me, although special-casing immediately nested conditionals seems safer and more sensible than temporarily blinding the instantiation depth limiter. I agree that it needs a bit more justification, either:

1. Example uses: (a) important packages with deeply immediately nested conditional types, or (b) a count of all immediately nested conditional types in some corpus like Definitely Typed or the user tests.
2. Performance benefit: by measuring some important package (or packages) before and after the change.

[weswigham]

@ahejlsberg I've finished merging #36584 into this PR, and also merged with master for you (it had conflicts). This should be good to go once CI checks all come back OK, so:

@typescript-bot test this
@typescript-bot user test this
@typescript-bot run dt
@typescript-bot perf test this

[typescript-bot]

Heya @weswigham, I've started to run the parallelized community code test suite on this PR at e59c3e6. You can monitor the build here. It should now contribute to this PR's status checks.

[typescript-bot]

Heya @weswigham, I've started to run the extended test suite on this PR at e59c3e6. You can monitor the build here. It should now contribute to this PR's status checks.

[typescript-bot]

Heya @weswigham, I've started to run the parallelized Definitely Typed test suite on this PR at e59c3e6. You can monitor the build here. It should now contribute to this PR's status checks.

[typescript-bot]

Heya @weswigham, I've started to run the perf test suite on this PR at e59c3e6. You can monitor the build here. It should now contribute to this PR's status checks.

Update: The results are in!

[typescript-bot]

@weswigham
The results of the perf run you requested are in!

Here they are:

Comparison Report - master..36583

Metric	master	36583	Delta	Best	Worst
Angular - node (v10.16.3, x64)
Memory used	334,095k (± 0.04%)	333,539k (± 0.02%)	-556k (- 0.17%)	333,407k	333,655k
Parse Time	1.62s (± 0.68%)	1.62s (± 0.55%)	-0.01s (- 0.49%)	1.60s	1.64s
Bind Time	0.89s (± 1.18%)	0.89s (± 1.15%)	+0.00s (+ 0.11%)	0.86s	0.91s
Check Time	4.69s (± 0.70%)	4.71s (± 0.70%)	+0.02s (+ 0.32%)	4.65s	4.82s
Emit Time	5.24s (± 0.57%)	5.23s (± 0.65%)	-0.01s (- 0.27%)	5.15s	5.32s
Total Time	12.45s (± 0.44%)	12.44s (± 0.52%)	-0.01s (- 0.04%)	12.29s	12.60s
Monaco - node (v10.16.3, x64)
Memory used	335,211k (± 0.02%)	335,194k (± 0.02%)	-18k (- 0.01%)	335,071k	335,272k
Parse Time	1.25s (± 0.98%)	1.25s (± 0.80%)	+0.00s (+ 0.40%)	1.23s	1.27s
Bind Time	0.78s (± 1.00%)	0.78s (± 0.75%)	-0.00s (- 0.26%)	0.77s	0.79s
Check Time	4.71s (± 0.88%)	4.71s (± 0.56%)	+0.00s (+ 0.04%)	4.66s	4.75s
Emit Time	2.92s (± 0.55%)	2.91s (± 0.72%)	-0.01s (- 0.45%)	2.87s	2.94s
Total Time	9.66s (± 0.67%)	9.65s (± 0.52%)	-0.01s (- 0.10%)	9.56s	9.74s
TFS - node (v10.16.3, x64)
Memory used	299,331k (± 0.02%)	299,322k (± 0.02%)	-9k (- 0.00%)	299,122k	299,468k
Parse Time	0.94s (± 0.59%)	0.95s (± 0.77%)	+0.01s (+ 1.07%)	0.93s	0.96s
Bind Time	0.75s (± 0.80%)	0.74s (± 0.67%)	-0.00s (- 0.00%)	0.74s	0.76s
Check Time	4.24s (± 0.56%)	4.27s (± 0.32%)	+0.03s (+ 0.61%)	4.23s	4.29s
Emit Time	3.01s (± 0.77%)	3.03s (± 0.68%)	+0.01s (+ 0.43%)	2.99s	3.08s
Total Time	8.94s (± 0.46%)	8.98s (± 0.40%)	+0.05s (+ 0.53%)	8.93s	9.08s
Angular - node (v12.1.0, x64)
Memory used	309,750k (± 0.03%)	309,174k (± 0.02%)	-576k (- 0.19%)	309,044k	309,264k
Parse Time	1.57s (± 0.64%)	1.57s (± 0.52%)	-0.00s (- 0.19%)	1.55s	1.59s
Bind Time	0.87s (± 1.13%)	0.87s (± 0.89%)	-0.00s (- 0.34%)	0.86s	0.89s
Check Time	4.59s (± 0.35%)	4.59s (± 0.51%)	-0.00s (- 0.00%)	4.51s	4.62s
Emit Time	5.40s (± 0.75%)	5.41s (± 0.80%)	+0.01s (+ 0.24%)	5.32s	5.52s
Total Time	12.43s (± 0.41%)	12.43s (± 0.48%)	+0.00s (+ 0.02%)	12.28s	12.59s
Monaco - node (v12.1.0, x64)
Memory used	315,128k (± 0.02%)	315,042k (± 0.02%)	-86k (- 0.03%)	314,869k	315,179k
Parse Time	1.20s (± 0.56%)	1.21s (± 0.96%)	+0.00s (+ 0.33%)	1.18s	1.22s
Bind Time	0.75s (± 0.99%)	0.74s (± 0.54%)	-0.01s (- 1.20%)	0.73s	0.75s
Check Time	4.54s (± 0.55%)	4.55s (± 0.64%)	+0.01s (+ 0.20%)	4.49s	4.61s
Emit Time	2.95s (± 1.09%)	2.97s (± 1.34%)	+0.01s (+ 0.44%)	2.91s	3.12s
Total Time	9.45s (± 0.61%)	9.47s (± 0.69%)	+0.02s (+ 0.19%)	9.35s	9.68s
TFS - node (v12.1.0, x64)
Memory used	281,603k (± 0.02%)	281,634k (± 0.02%)	+31k (+ 0.01%)	281,524k	281,751k
Parse Time	0.92s (± 0.81%)	0.92s (± 0.63%)	-0.00s (- 0.00%)	0.91s	0.93s
Bind Time	0.71s (± 0.73%)	0.71s (± 0.94%)	+0.00s (+ 0.14%)	0.70s	0.72s
Check Time	4.18s (± 0.45%)	4.18s (± 0.53%)	-0.00s (- 0.05%)	4.13s	4.25s
Emit Time	3.05s (± 0.99%)	3.06s (± 1.30%)	+0.01s (+ 0.49%)	2.98s	3.18s
Total Time	8.86s (± 0.58%)	8.88s (± 0.45%)	+0.02s (+ 0.19%)	8.80s	8.99s
Angular - node (v8.9.0, x64)
Memory used	328,956k (± 0.02%)	328,449k (± 0.01%)	-508k (- 0.15%)	328,352k	328,508k
Parse Time	2.10s (± 0.57%)	2.10s (± 0.53%)	-0.00s (- 0.05%)	2.08s	2.13s
Bind Time	0.92s (± 1.03%)	0.92s (± 0.70%)	+0.00s (+ 0.55%)	0.91s	0.94s
Check Time	5.47s (± 0.59%)	5.46s (± 0.71%)	-0.01s (- 0.16%)	5.37s	5.55s
Emit Time	6.26s (± 0.98%)	6.22s (± 0.91%)	-0.04s (- 0.64%)	6.14s	6.37s
Total Time	14.76s (± 0.50%)	14.71s (± 0.52%)	-0.05s (- 0.30%)	14.60s	14.93s
Monaco - node (v8.9.0, x64)
Memory used	333,388k (± 0.02%)	333,429k (± 0.01%)	+41k (+ 0.01%)	333,357k	333,505k
Parse Time	1.54s (± 0.32%)	1.53s (± 0.32%)	-0.00s (- 0.13%)	1.52s	1.54s
Bind Time	0.91s (± 0.75%)	0.90s (± 0.68%)	-0.01s (- 0.55%)	0.89s	0.91s
Check Time	5.38s (± 0.82%)	5.39s (± 0.76%)	+0.01s (+ 0.28%)	5.30s	5.52s
Emit Time	3.55s (± 0.71%)	3.54s (± 0.35%)	-0.01s (- 0.34%)	3.52s	3.57s
Total Time	11.37s (± 0.57%)	11.37s (± 0.44%)	-0.00s (- 0.00%)	11.26s	11.52s
TFS - node (v8.9.0, x64)
Memory used	298,693k (± 0.01%)	298,703k (± 0.02%)	+10k (+ 0.00%)	298,586k	298,813k
Parse Time	1.25s (± 0.48%)	1.25s (± 0.49%)	-0.00s (- 0.32%)	1.24s	1.27s
Bind Time	0.75s (± 0.53%)	0.75s (± 0.99%)	+0.00s (+ 0.53%)	0.74s	0.77s
Check Time	4.80s (± 0.51%)	4.83s (± 0.52%)	+0.03s (+ 0.56%)	4.76s	4.86s
Emit Time	3.38s (± 0.72%)	3.39s (± 0.86%)	+0.01s (+ 0.27%)	3.30s	3.44s
Total Time	10.18s (± 0.41%)	10.22s (± 0.44%)	+0.04s (+ 0.38%)	10.10s	10.29s
Angular - node (v8.9.0, x86)
Memory used	188,662k (± 0.03%)	188,448k (± 0.02%)	-214k (- 0.11%)	188,370k	188,521k
Parse Time	2.05s (± 0.59%)	2.06s (± 0.74%)	+0.01s (+ 0.34%)	2.02s	2.09s
Bind Time	1.07s (± 0.47%)	1.07s (± 0.46%)	+0.01s (+ 0.85%)	1.06s	1.08s
Check Time	5.01s (± 0.53%)	4.99s (± 0.51%)	-0.01s (- 0.30%)	4.93s	5.05s
Emit Time	6.13s (± 0.89%)	6.17s (± 1.04%)	+0.04s (+ 0.67%)	6.07s	6.35s
Total Time	14.25s (± 0.53%)	14.29s (± 0.61%)	+0.05s (+ 0.34%)	14.12s	14.53s
Monaco - node (v8.9.0, x86)
Memory used	189,091k (± 0.01%)	189,118k (± 0.02%)	+27k (+ 0.01%)	189,056k	189,225k
Parse Time	1.59s (± 1.00%)	1.59s (± 0.42%)	-0.00s (- 0.13%)	1.58s	1.61s
Bind Time	0.77s (± 0.80%)	0.77s (± 0.91%)	+0.00s (+ 0.26%)	0.76s	0.79s
Check Time	5.30s (± 2.36%)	5.36s (± 1.93%)	+0.06s (+ 1.17%)	5.10s	5.53s
Emit Time	3.00s (± 4.15%)	2.99s (± 3.51%)	-0.02s (- 0.53%)	2.83s	3.21s
Total Time	10.66s (± 0.48%)	10.71s (± 0.40%)	+0.05s (+ 0.42%)	10.60s	10.80s
TFS - node (v8.9.0, x86)
Memory used	170,280k (± 0.01%)	170,286k (± 0.02%)	+7k (+ 0.00%)	170,172k	170,379k
Parse Time	1.28s (± 1.16%)	1.27s (± 0.54%)	-0.00s (- 0.23%)	1.26s	1.29s
Bind Time	0.71s (± 0.48%)	0.72s (± 0.83%)	+0.00s (+ 0.42%)	0.71s	0.73s
Check Time	4.60s (± 1.01%)	4.59s (± 0.71%)	-0.01s (- 0.11%)	4.52s	4.66s
Emit Time	2.97s (± 0.98%)	2.96s (± 0.79%)	-0.01s (- 0.34%)	2.90s	3.01s
Total Time	9.55s (± 0.64%)	9.54s (± 0.41%)	-0.01s (- 0.13%)	9.48s	9.67s

System

Machine Name	ts-ci-ubuntu
Platform	linux 4.4.0-166-generic
Architecture	x64
Available Memory	16 GB
Available Memory	1 GB
CPUs	4 × Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz

Hosts

• node (v10.16.3, x64)
• node (v12.1.0, x64)
• node (v8.9.0, x64)
• node (v8.9.0, x86)

Scenarios

• Angular - node (v10.16.3, x64)
• Angular - node (v12.1.0, x64)
• Angular - node (v8.9.0, x64)
• Angular - node (v8.9.0, x86)
• Monaco - node (v10.16.3, x64)
• Monaco - node (v12.1.0, x64)
• Monaco - node (v8.9.0, x64)
• Monaco - node (v8.9.0, x86)
• TFS - node (v10.16.3, x64)
• TFS - node (v12.1.0, x64)
• TFS - node (v8.9.0, x64)
• TFS - node (v8.9.0, x86)

Benchmark	Name	Iterations
Current	36583	10
Baseline	master	10

[sandersn]

@ahejlsberg This looks ready to merge, right?

[sandersn]

@ahejlsberg can you update this PR and then merge it?