Fix #3965: Trying to make higher-kinded equality correct and efficient [WIP] #3975
Conversation
Since correct hashing under binders seems to be very expensive (see performance data for scala#3970), let's try have fewer types that require this.
odersky
changed the title
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 0 running. |
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (39980f3) |
1 similar comment
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (39980f3) |
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 0 running. |
Rework hashing and equality so that two isomorphic types are identified even if they are dependent (i.e. have back edges from a BoundType such as ParamRef or RecThis to its HKLambda or RecType Method and PolyTypes are still generative. This fixes scala#3965
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (39980f3) |
Before hashing with a non-null binders list, check whether the type has a stable hash. Hash-stability results are cached in NamedTypes and AppliedTypes.
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (39980f3) |
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (39980f3) |
1 similar comment
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (39980f3) |
When compiling dotc/typer/*.scala we observe a reduction of the number of calls from ~3.7m to ~1.0m.
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 0 running. |
|
test performance please |
|
performance test scheduled: 1 job(s) in queue, 1 running. |
|
@liufengyun @allanrenucci Is there a timeout for performance tests? My second to last test contained a bug that makes it go basically forever. If there's no timeout we need to abort this so that other tests can run. |
|
@odersky Now it's killed, the other is running. |
|
@liufengyun Thanks! |
|
Performance test finished successfully: Visit http://dotty-bench.epfl.ch/3975/ to see the changes. Benchmarks is based on merging with master (5943331) |
|
Subsumed by #3978 |
#3970 fixed equality for higher-kinded types but came at a steep performance penalty. This PR is trying to avoid the penalty.
First, since correct hashing under binders seems to be very expensive (see performance
data for #3970), let's try have fewer types that require this.