Dan Moseley
Dan Moseley
I'm not so sure it won't be measurable- we have seen such things regress hot paths (which this is) eg by making code longer so it doesn't inline any more....
There are various rules and heuristics, and it may depend on the caller. Cc @AndyAyersMS for thoughts
@stephentoub what is your opinion on the value of this change?
We used to intern some hard coded strings like true and false in MSBuild including in that codepath ("OpportunisticIntern"). I can't look at the code at the moment but I...
https://github.com/dotnet/msbuild/pull/5663/files#diff-8b918fc3cf86a6bcec4ce6b6c0adce1f61dd3bb09b19c61320cf6fb51aa0ec76L331
Any of these worth a breaking change note? (I'm guessing not)
Test failure: ConsoleEncoding.TestEncoding - NotSupportedException : Memory stream is not expandable
When test hits this exception it should log the code page, input encoding and BOM. It's quite curious it's sporadic and on random strings. Something non deterministic/racy.
@Zintom why do you think a heap allocator would be cheaper? How would the GC detect roots from your heap?
@roji Is this destined for .NET 7? The cutoff is almost upon us so wondering who owns next action.
@aortiz-msft