StringMap always copies its strings into its own storage. A DenseMap
of StringRefs has the same caveats as any other use of StringRef, but
in the cases I've changed the string has very clear ownership that
outlives the map.
No functionality change, but should reduce memory usage and malloc
traffic a little.
When provided, this flag warns about implicit overrides, where a
declaration overrides another declaration but is not marked with the
‘override’ keyword. The warning can be suppressed by either providing
‘override’ or ‘@_nonoverride’.
At present, this only happens with overrides in protocols.
* [InterfaceGen] Remove #ifs from default args
This patch removes all #if configs form the bodies of default arguments,
which can contain multiline closures, while preserving the bodies of the
clauses that are active.
This code is generalized and should "just work" for inlinable function
bodies, which will come in a later patch.
* Address review comments
* Fix and test CharSourceRange.overlaps
* Fix CharSourceRange::print to respect half-open ranges
The order of evaluation for arguments is underspecified in the language
specification. It is permissible to evluate right to left, resulting in the use
after the move of the OwnedPtr. This happens in practice on Windows. Alter the
construction to avoid this.
ExponentialGrowthAppendingBinaryByteStream::writeRaw() is sometimes used for unaligned writes, so it shouldn’t be casting and dereferencing pointers. Tests in Godbolt indicate that the call to memcpy() is optimized into similar instructions anyway.
And tidy up doc comments for other methods that do or don't respect
'#sourceLocation'.
'getBufferIdentifierForLoc' is no longer useful: it doesn't
consistently return either the name of an actual buffer or the name of
a file suitable for diagnostics. As seen in the previous commit, all
remaining uses of it were wrong anyway. Remove it.
We refactor the existing USR generation code to use the request evaluator
model. Most part of this patch is NFC, however a slight difference is
using this model allows us to cache the calculated USRs in the evaluator.
If we know the size of a type at compile time (like we do for all the
integer types), it is cheaper to assign the data buffer directly instead
of using a memcpy.
It is more efficient than llvm::AppendingBinaryByteStream if a lot of
small data gets appended to it because it doesn't need to resize its
buffer on each write.
We cannot use unowned strings for token texts of incrementally parsed
syntax trees since the source buffer to which reused nodes refer will
have been freed for reused nodes. Always copying the token text whenever
OwnedString is passed is too expensive. A reference counted copy of the
string allows us to keep the token's string alive across incremental
parses while eliminating unnecessary copies.
Having this be a single buffer hardcoded in the SourceManager and set
by all clients is silly. SourceFiles with the 'Main' kind are allowed
to have hashbang lines (`#!`), other files are not. And anyone
manually setting up a Lexer can decide for themselves.
No intended behavioral change.
This is helpful in experimenting with constraint solver changes that
might help us remove some of these unsound options. It's not ever mean
to be enabled, but if we're able to remove the things guarded by the
option we can eventually remove the option.
