Fix CFA for with generic T (#62133) #62151

kaushik-i-b · 2025-07-30T22:35:46Z

Fixes

Fixes #62133 – x && x should not narrow generic T to NonNullable<T>.

What changed

Modified getDefinitelyFalsyPartOfType in src/compiler/checker.ts
- Added a branch for TypeParameter that consults the type’s constraint
- Added recursion guard to avoid infinite loops

New behaviour

function f<T>(x: T) {
    return x && x;   // now typed as T, not NonNullable<T>
}

kaushik-i-b · 2025-07-30T23:24:30Z

@microsoft-github-policy-service agree

RyanCavanaugh · 2025-07-31T16:32:29Z

@typescript-bot test it

typescript-bot · 2025-07-31T16:32:35Z

Starting jobs; this comment will be updated as builds start and complete.

Command	Status	Results
`test top400`	✅ Started	✅ Results
`user test this`	✅ Started	✅ Results
`run dt`	✅ Started	✅ Results
`perf test this faster`	✅ Started	👀 Results

typescript-bot · 2025-07-31T16:55:49Z

Hey @RyanCavanaugh, the results of running the DT tests are ready.

Everything looks the same!

You can check the log here.

typescript-bot · 2025-07-31T17:01:01Z

@RyanCavanaugh Here are the results of running the user tests with tsc comparing main and refs/pull/62151/merge:

Everything looks good!

typescript-bot · 2025-07-31T17:18:48Z

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

Here they are:

tsc

Comparison Report - baseline..pr

_Metric	_baseline	_pr	_Delta	_Best	_Worst	_p-value
_{Compiler-Unions - node (v18.15.0, x64)}
_Errors	₃₄	₃₄	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_62,370	_62,370	_~	_~	_~	_{p=1.000 n=6}
_Types	_50,386	_50,386	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{194,192k (± 0.94%)}	_{194,132k (± 0.96%)}	_~	_192,910k	_196,546k	_{p=1.000 n=6}
_{Parse Time}	_{1.31s (± 0.62%)}	_{1.32s (± 0.80%)}	_~	_1.30s	_1.33s	_{p=0.180 n=6}
_{Bind Time}	_0.73s	_0.73s	_~	_~	_~	_{p=1.000 n=6}
_{Check Time}	_{9.75s (± 0.08%)}	_{9.74s (± 0.32%)}	_~	_9.68s	_9.76s	_{p=0.446 n=6}
_{Emit Time}	_{2.74s (± 0.83%)}	_{2.75s (± 0.27%)}	_~	_2.74s	_2.76s	_{p=0.741 n=6}
_{Total Time}	_{14.52s (± 0.19%)}	_{14.53s (± 0.20%)}	_~	_14.47s	_14.55s	_{p=0.744 n=6}
_{angular-1 - node (v18.15.0, x64)}
_Errors	₅₆	₅₆	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_948,914	_948,914	_~	_~	_~	_{p=1.000 n=6}
_Types	_410,884	_410,884	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{1,226,424k (± 0.01%)}	_{1,226,352k (± 0.01%)}	_~	_1,226,278k	_1,226,431k	_{p=0.093 n=6}
_{Parse Time}	_{6.52s (± 0.84%)}	_{6.50s (± 0.44%)}	_~	_6.47s	_6.54s	_{p=1.000 n=6}
_{Bind Time}	_{1.87s (± 0.22%)}	_{1.87s (± 0.28%)}	_~	_1.87s	_1.88s	_{p=0.595 n=6}
_{Check Time}	_{31.99s (± 0.37%)}	_{32.01s (± 0.27%)}	_~	_31.91s	_32.15s	_{p=1.000 n=6}
_{Emit Time}	_{14.89s (± 0.19%)}	_{14.83s (± 0.61%)}	_~	_14.66s	_14.89s	_{p=0.166 n=6}
_{Total Time}	_{55.26s (± 0.28%)}	_{55.21s (± 0.28%)}	_~	_54.95s	_55.41s	_{p=0.423 n=6}
_{mui-docs - node (v18.15.0, x64)}
_Errors	₀	₀	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_2,543,160	_2,543,160	_~	_~	_~	_{p=1.000 n=6}
_Types	_900,384	_900,384	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{2,829,584k (± 0.01%)}	_{2,829,471k (± 0.01%)}	_~	_2,829,258k	_2,829,674k	_{p=0.230 n=6}
_{Parse Time}	_{8.76s (± 0.38%)}	_{8.76s (± 0.26%)}	_~	_8.73s	_8.79s	_{p=0.936 n=6}
_{Bind Time}	_{2.25s (± 0.52%)}	_{2.26s (± 0.65%)}	_~	_2.24s	_2.28s	_{p=0.142 n=6}
_{Check Time}	_{85.71s (± 0.27%)}	_{85.64s (± 0.40%)}	_~	_85.35s	_86.12s	_{p=0.810 n=6}
_{Emit Time}	_{0.93s (±103.79%)}	_{1.00s (±107.26%)}	_~	_0.30s	_2.41s	_{p=0.797 n=6}
_{Total Time}	_{97.65s (± 1.03%)}	_{97.66s (± 0.94%)}	_~	_96.69s	_98.83s	_{p=1.000 n=6}
_{self-build-src - node (v18.15.0, x64)}
_Errors	₀	₂	_{🔻+2 (+ ∞%)}	_~	_~	_{p=0.001 n=6}
_Symbols	_1,227,065	_1,301,666	_{🔻+74,601 (+ 6.08%)}	_~	_~	_{p=0.001 n=6}
_Types	_267,478	_267,573	_{+95 (+ 0.04%)}	_~	_~	_{p=0.001 n=6}
_{Memory used}	_{2,423,713k (± 6.15%)}	_{2,584,364k (± 5.76%)}	_{🔻+160,652k (+ 6.63%)}	_2,522,695k	_2,888,318k	_{p=0.045 n=6}
_{Parse Time}	_{5.22s (± 0.92%)}	_{6.56s (± 1.08%)}	_{🔻+1.34s (+25.65%)}	_6.48s	_6.65s	_{p=0.005 n=6}
_{Bind Time}	_{1.78s (± 0.90%)}	_{2.42s (± 1.12%)}	_{🔻+0.64s (+35.99%)}	_2.39s	_2.46s	_{p=0.004 n=6}
_{Check Time}	_{35.30s (± 0.51%)}	_{35.27s (± 0.46%)}	_~	_35.04s	_35.54s	_{p=0.575 n=6}
_{Emit Time}	_{3.04s (± 4.61%)}	_{3.00s (± 0.81%)}	_~	_2.97s	_3.03s	_{p=0.936 n=6}
_{Total Time}	_{45.36s (± 0.59%)}	_{47.25s (± 0.29%)}	_{🔻+1.89s (+ 4.18%)}	_47.06s	_47.44s	_{p=0.005 n=6}
_{self-build-src-public-api - node (v18.15.0, x64)}
_Errors	₀	₂	_{🔻+2 (+ ∞%)}	_~	_~	_{p=0.001 n=6}
_Symbols	_1,227,065	_1,301,666	_{🔻+74,601 (+ 6.08%)}	_~	_~	_{p=0.001 n=6}
_Types	_267,478	_267,573	_{+95 (+ 0.04%)}	_~	_~	_{p=0.001 n=6}
_{Memory used}	_{2,919,396k (±12.88%)}	_{2,963,039k (±13.49%)}	_~	_2,597,311k	_3,328,976k	_{p=0.378 n=6}
_{Parse Time}	_{8.49s (± 1.45%)}	_{10.65s (± 1.46%)}	_{🔻+2.16s (+25.48%)}	_10.46s	_10.82s	_{p=0.005 n=6}
_{Bind Time}	_{2.66s (± 1.51%)}	_{3.59s (± 1.36%)}	_{🔻+0.93s (+34.77%)}	_3.52s	_3.63s	_{p=0.005 n=6}
_{Check Time}	_{53.19s (± 0.45%)}	_{53.31s (± 0.38%)}	_~	_53.09s	_53.55s	_{p=0.230 n=6}
_{Emit Time}	_{4.48s (± 3.53%)}	_{4.47s (± 4.18%)}	_~	_4.24s	_4.70s	_{p=0.810 n=6}
_{Total Time}	_{68.82s (± 0.25%)}	_{72.03s (± 0.26%)}	_{🔻+3.21s (+ 4.66%)}	_71.75s	_72.25s	_{p=0.005 n=6}
_{self-compiler - node (v18.15.0, x64)}
_Errors	₀	₀	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_262,554	_262,555	_{+1 (+ 0.00%)}	_~	_~	_{p=0.001 n=6}
_Types	_107,163	_107,163	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{441,960k (± 0.02%)}	_{441,953k (± 0.02%)}	_~	_441,856k	_442,096k	_{p=0.689 n=6}
_{Parse Time}	_{3.56s (± 0.77%)}	_{3.52s (± 0.95%)}	_~	_3.47s	_3.55s	_{p=0.089 n=6}
_{Bind Time}	_{1.31s (± 0.96%)}	_{1.33s (± 0.95%)}	_{+0.02s (+ 1.53%)}	_1.31s	_1.34s	_{p=0.034 n=6}
_{Check Time}	_{18.90s (± 0.38%)}	_{18.93s (± 0.26%)}	_~	_18.86s	_19.00s	_{p=0.810 n=6}
_{Emit Time}	_{1.53s (± 0.87%)}	_{1.52s (± 1.46%)}	_~	_1.49s	_1.55s	_{p=0.566 n=6}
_{Total Time}	_{25.30s (± 0.34%)}	_{25.29s (± 0.24%)}	_~	_25.20s	_25.37s	_{p=0.810 n=6}
_{ts-pre-modules - node (v18.15.0, x64)}
_Errors	₇₁	₇₁	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_225,367	_225,367	_~	_~	_~	_{p=1.000 n=6}
_Types	_94,290	_94,290	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{371,172k (± 0.02%)}	_{371,143k (± 0.02%)}	_~	_371,075k	_371,229k	_{p=0.575 n=6}
_{Parse Time}	_{2.88s (± 0.75%)}	_{2.88s (± 1.34%)}	_~	_2.83s	_2.92s	_{p=0.627 n=6}
_{Bind Time}	_{1.59s (± 1.42%)}	_{1.59s (± 0.83%)}	_~	_1.57s	_1.61s	_{p=0.870 n=6}
_{Check Time}	_{16.46s (± 0.31%)}	_{16.43s (± 0.33%)}	_~	_16.38s	_16.52s	_{p=0.227 n=6}
_{Emit Time}	_{0.00s (±244.70%)}	_0.00s	_~	_~	_~	_{p=0.405 n=6}
_{Total Time}	_{20.93s (± 0.13%)}	_{20.90s (± 0.37%)}	_~	_20.81s	_21.04s	_{p=0.149 n=6}
_{vscode - node (v18.15.0, x64)}
_Errors	₁	₁	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_3,559,337	_3,559,337	_~	_~	_~	_{p=1.000 n=6}
_Types	_1,200,047	_1,200,047	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{3,603,620k (± 0.00%)}	_{3,603,664k (± 0.00%)}	_~	_3,603,514k	_3,603,767k	_{p=0.575 n=6}
_{Parse Time}	_{15.24s (± 0.60%)}	_{15.26s (± 0.70%)}	_~	_15.16s	_15.43s	_{p=0.936 n=6}
_{Bind Time}	_{4.91s (± 0.26%)}	_{4.92s (± 0.61%)}	_~	_4.89s	_4.96s	_{p=0.678 n=6}
_{Check Time}	_{99.86s (± 1.16%)}	_{100.26s (± 1.20%)}	_~	_99.60s	_102.69s	_{p=0.378 n=6}
_{Emit Time}	_{31.75s (± 8.60%)}	_{30.84s (± 0.98%)}	_~	_30.34s	_31.14s	_{p=0.936 n=6}
_{Total Time}	_{151.77s (± 2.16%)}	_{151.28s (± 0.76%)}	_~	_150.29s	_153.54s	_{p=0.689 n=6}
_{webpack - node (v18.15.0, x64)}
_Errors	₂	₂	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_321,136	_321,136	_~	_~	_~	_{p=1.000 n=6}
_Types	_140,173	_140,173	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{477,420k (± 0.03%)}	_{477,487k (± 0.02%)}	_~	_477,363k	_477,606k	_{p=0.575 n=6}
_{Parse Time}	_{4.28s (± 0.57%)}	_{4.27s (± 0.32%)}	_~	_4.25s	_4.28s	_{p=0.285 n=6}
_{Bind Time}	_{1.79s (± 1.24%)}	_{1.80s (± 0.93%)}	_~	_1.78s	_1.82s	_{p=0.512 n=6}
_{Check Time}	_{20.47s (± 0.35%)}	_{20.54s (± 0.39%)}	_~	_20.46s	_20.69s	_{p=0.229 n=6}
_{Emit Time}	_{0.00s (±154.76%)}	_{0.00s (±244.70%)}	_~	_0.00s	_0.01s	_{p=0.595 n=6}
_{Total Time}	_{26.55s (± 0.31%)}	_{26.61s (± 0.32%)}	_~	_26.54s	_26.77s	_{p=0.229 n=6}
_{xstate-main - node (v18.15.0, x64)}
_Errors	₃₀	₃₀	_~	_~	_~	_{p=1.000 n=6}
_Symbols	_663,079	_663,079	_~	_~	_~	_{p=1.000 n=6}
_Types	_197,935	_197,935	_~	_~	_~	_{p=1.000 n=6}
_{Memory used}	_{570,047k (± 0.01%)}	_{570,082k (± 0.01%)}	_~	_569,994k	_570,145k	_{p=0.470 n=6}
_{Parse Time}	_{4.28s (± 0.80%)}	_{4.30s (± 0.49%)}	_~	_4.26s	_4.32s	_{p=0.332 n=6}
_{Bind Time}	_{1.33s (± 1.04%)}	_{1.34s (± 1.02%)}	_~	_1.32s	_1.36s	_{p=0.249 n=6}
_{Check Time}	_{20.00s (± 1.52%)}	_{20.00s (± 1.33%)}	_~	_19.81s	_20.52s	_{p=0.689 n=6}
_{Emit Time}	_{0.00s (±244.70%)}	_0.00s	_~	_~	_~	_{p=0.405 n=6}
_{Total Time}	_{25.61s (± 1.06%)}	_{25.64s (± 1.10%)}	_~	_25.44s	_26.19s	_{p=0.630 n=6}

System info unknown

Hosts

node (v18.15.0, x64)

Scenarios

Compiler-Unions - node (v18.15.0, x64)
angular-1 - node (v18.15.0, x64)
mui-docs - node (v18.15.0, x64)
self-build-src - node (v18.15.0, x64)
self-build-src-public-api - node (v18.15.0, x64)
self-compiler - node (v18.15.0, x64)
ts-pre-modules - node (v18.15.0, x64)
vscode - node (v18.15.0, x64)
webpack - node (v18.15.0, x64)
xstate-main - node (v18.15.0, x64)

_Benchmark	_Name	_Iterations
_Current	_pr	₆
_Baseline	_baseline	₆

Developer Information:

Download Benchmarks

typescript-bot · 2025-07-31T18:16:07Z

@RyanCavanaugh Here are the results of running the top 400 repos with tsc comparing main and refs/pull/62151/merge:

Everything looks good!

kaushik-i-b · 2025-08-01T09:59:07Z

Will my PR be merged.? 😅

RyanCavanaugh · 2025-08-01T15:25:04Z

What's the non-degenerate repro where this matters?

kaushik-i-b · 2025-08-01T16:19:38Z

I pushed commit that removes the new test file and instead adds a minimal, non-degenerate test into an existing file:

New case:

// Generic type parameter keeps falsy constituents (#62133)
declare function id<T>(x: T): T;
function f<T extends string | null>(x: T) {
    const y = x && x;   // should be T, not NonNullable<T>
    return y;
}
const t1 = f("a");   // "a"
const t2 = f(null);  // null  (was incorrectly never)

The baseline diff now shows only the expected new lines, and the
change is exercised without a dedicated file.

RyanCavanaugh · 2025-08-01T18:43:00Z

There's no reason to write x && x, though. What's code that you would write on purpose that doesn't work without this PR?

kaushik-i-b · 2025-08-01T18:49:55Z

There's no reason to write x && x, though. What's code that you would write on purpose that doesn't work without this PR?

Sorry for the confusion...😅
Code fix explained in one sentence
We taught getDefinitelyFalsyPartOfType to look at the constraint of a generic type parameter instead of immediately returning never, so CFA no longer narrows T && anything to NonNullable<T>.

What the old code did

if (type.flags & TypeFlags.TypeParameter) {
    return neverType;   // “the falsy part of T is nothing”
}

Result: in control-flow analysis the expression left && right is calculated as
left minus nothing ⇒ left becomes NonNullable<T>.

What the new code does

if (type.flags & TypeFlags.TypeParameter) {
    const constraint = getBaseConstraintOfType(type);
    if (!constraint || constraint === type) return neverType;
    return getDefinitelyFalsyPartOfType(constraint);
}

If T extends string | null, its constraint is string | null;
getDefinitelyFalsyPartOfType(string | null) correctly returns null;
CFA now subtracts only null from T, leaving the original union intact.
No NonNullable narrowing, no runtime surprises.

RyanCavanaugh · 2025-08-01T18:53:45Z

What user code is affected by this in a situation where the code makes sense in the first place? How can I observe if this PR is present or not without writing nonsense code like x && x ?

If there's not an answer to that question, I don't think we should merge this -- code changes need to be well-motivated by actual examples.

kaushik-i-b · 2025-08-01T19:05:11Z

What user code is affected by this in a situation where the code makes sense in the first place? How can I observe if this PR is present or not without writing nonsense code like x && x ?

If there's not an answer to that question, I don't think we should merge this -- code changes need to be well-motivated by actual examples.

After the discussion above I agree the real-world impact is too small to justify the extra complexity. I’m closing the PR—thanks for the thorough review!
Thank you

Fix CFA for with generic T (microsoft#62133)

3ee0907

github-project-automation bot added this to PR Backlog Jul 30, 2025

github-project-automation bot moved this to Not started in PR Backlog Jul 30, 2025

typescript-bot added the For Uncommitted Bug PR for untriaged, rejected, closed or missing bug label Jul 30, 2025

kaushik-i-b added 2 commits July 31, 2025 03:01

Update package.json

8ac8fcd

Update package.json

a07c16a

kaushik-i-b closed this Aug 1, 2025

github-project-automation bot moved this from Not started to Done in PR Backlog Aug 1, 2025

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Fix CFA for with generic T (#62133) #62151

Fix CFA for with generic T (#62133) #62151

Uh oh!

kaushik-i-b commented Jul 30, 2025

Uh oh!

kaushik-i-b commented Jul 30, 2025

Uh oh!

RyanCavanaugh commented Jul 31, 2025

Uh oh!

typescript-bot commented Jul 31, 2025 •

edited

Loading

Uh oh!

typescript-bot commented Jul 31, 2025

Uh oh!

typescript-bot commented Jul 31, 2025

Uh oh!

typescript-bot commented Jul 31, 2025

tsc

Uh oh!

typescript-bot commented Jul 31, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025 •

edited

Loading

Uh oh!

RyanCavanaugh commented Aug 1, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025 •

edited

Loading

Uh oh!

RyanCavanaugh commented Aug 1, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025

Uh oh!

RyanCavanaugh commented Aug 1, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025

Uh oh!

Uh oh!

Fix CFA for with generic T (#62133) #62151

Fix CFA for with generic T (#62133) #62151

Uh oh!

Conversation

kaushik-i-b commented Jul 30, 2025

Fixes

What changed

New behaviour

Uh oh!

kaushik-i-b commented Jul 30, 2025

Uh oh!

RyanCavanaugh commented Jul 31, 2025

Uh oh!

typescript-bot commented Jul 31, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

typescript-bot commented Jul 31, 2025

Uh oh!

typescript-bot commented Jul 31, 2025

Uh oh!

typescript-bot commented Jul 31, 2025

tsc

Uh oh!

typescript-bot commented Jul 31, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RyanCavanaugh commented Aug 1, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

RyanCavanaugh commented Aug 1, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025

Uh oh!

RyanCavanaugh commented Aug 1, 2025

Uh oh!

kaushik-i-b commented Aug 1, 2025

Uh oh!

Uh oh!

typescript-bot commented Jul 31, 2025 •

edited

Loading

kaushik-i-b commented Aug 1, 2025 •

edited

Loading

kaushik-i-b commented Aug 1, 2025 •

edited

Loading