Add error reporting when looking up types by demangled name.

[mikeash]

When mangled name lookups fail, it can be pretty hard to track down the cause. There are many places where it can fail, and when nested generics are involved you can end up recursing a lot. My technique so far has involved adding lots and lots of ad-hoc logging calls to track down the failure.

This PR plumbs error reporting through the whole structure. Errors are recorded as a function pointer and a context pointer, allowing errors to gather relevant information, but also avoiding the overhead of string building on paths where the caller is going to have some sort of fallback instead of reporting the error.

For an example of what this does, here's an error I debugged the other day:

failed to demangle witness for associated type 'Body' in conformance 'Syndrome.ProcessPaymentButton: View' from mangled name '+�'

And here's what that error looks like with this PR:

failed to demangle witness for associated type 'Body' in conformance 'Syndrome.ProcessPaymentButton: View' from mangled name '+�' - _gatherGenericParameters: context: $s7SwiftUI12ProgressViewVMn 0x1a214efd0 <SwiftUI.EmptyView> parent: <null> - incorrect number of generic args (1), 2 local params, 2 total params

This would have saved a huge amount of debugging time (and hopefully will save a huge amount in the future!).

[mikeash]

@swift-ci please benchmark

[swift-ci]

Build failed before running benchmark.

[mikeash]

@swift-ci please benchmark

[swift-ci]

Build failed before running benchmark.

[mikeash]

@swift-ci please test

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - 0854fb609e77b7dab999a019d75f6df2e1f50b96

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - 0854fb609e77b7dab999a019d75f6df2e1f50b96

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please test

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - 0854fb609e77b7dab999a019d75f6df2e1f50b96

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please clean test

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - 0854fb609e77b7dab999a019d75f6df2e1f50b96

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - 0854fb609e77b7dab999a019d75f6df2e1f50b96

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please test

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - 1dec95205a759db9b81ca42d6e2a52893f190633

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - 1dec95205a759db9b81ca42d6e2a52893f190633

[tbkka]

"ErrorOr" happens to be my personal favorite name for this pattern. 😄

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please benchmark

[swift-ci]

Build failed before running benchmark.

[shahmishal]

swiftlang/llvm-project#1702

@swift-ci please benchmark

[swift-ci]

Build failed before running benchmark.

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please test

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please benchmark

[swift-ci]

Build failed
Swift Test Linux Platform
Git Sha - aa800c67906b49885e167a8bb6a7e1d9de382d60

[swift-ci]

Build failed
Swift Test OS X Platform
Git Sha - aa800c67906b49885e167a8bb6a7e1d9de382d60

[mikeash]

swiftlang/llvm-project#1702

@swift-ci please benchmark

[swift-ci]

Performance: -O

Improvement	OLD	NEW	DELTA	RATIO
Set.isSubset.Int.Empty	51	43	-15.7%	1.19x
Set.isStrictSubset.Int.Empty	51	45	-11.8%	1.13x (?)
Set.isSuperset.Seq.Empty.Int	50	45	-10.0%	1.11x (?)

Code size: -O

Performance: -Osize

Regression	OLD	NEW	DELTA	RATIO
ObjectiveCBridgeStubToNSStringRef	70	79	+12.9%	0.89x (?)
RemoveWhereFilterInts	24	27	+12.5%	0.89x
Improvement	OLD	NEW	DELTA	RATIO
FlattenListFlatMap	4535	3627	-20.0%	1.25x (?)
Set.isStrictSubset.Int.Empty	54	47	-13.0%	1.15x (?)
Set.isDisjoint.Seq.Int.Empty	51	47	-7.8%	1.09x (?)

Code size: -Osize

Performance: -Onone

Code size: -swiftlibs

How to read the data

The tables contain differences in performance which are larger than 8% and differences in code size which are larger than 1%.

If you see any unexpected regressions, you should consider fixing the
regressions before you merge the PR.

Noise: Sometimes the performance results (not code size!) contain false
alarms. Unexpected regressions which are marked with '(?)' are probably noise.
If you see regressions which you cannot explain you can try to run the
benchmarks again. If regressions still show up, please consult with the
performance team (@eeckstein).

Hardware Overview

  Model Name: Mac mini
  Model Identifier: Macmini8,1
  Processor Name: 6-Core Intel Core i7
  Processor Speed: 3.2 GHz
  Number of Processors: 1
  Total Number of Cores: 6
  L2 Cache (per Core): 256 KB
  L3 Cache: 12 MB
  Memory: 64 GB