Skip to content

Joe Groff
jckarter

Organizations

@apple
@jckarter
  • @jckarter 46f9c5d
    mark Unicode shims used in non-ObjC build
jckarter commented on commit apple/swift@668fbf8df9
@jckarter

@gribozavr Look OK now?

jckarter commented on pull request apple/swift#1217
@jckarter

@swift-ci Please test

@jckarter
jckarter commented on pull request apple/swift#1217
@jckarter

Yes, thanks!

jckarter commented on pull request apple/swift#1217
@jckarter

@swift-ci Please test

jckarter opened pull request apple/swift#1217
@jckarter
Runtime: Build with -fvisibility=hidden.
1 commit with 384 additions and 12 deletions
@jckarter
@jckarter

For non-template members of dependent superclasses, you can do using Super<T>::foo; or using typename Super<T>::Bar; to avoid the need for this-> e…

@jckarter

Dependent lookup can't know getTrailingObjects is a template, so the next < would be parsed as an operator without the template before it.

@jckarter
  • @jckarter fe52ef0
    Start review for SE-0029 removing implicit tuple splat
jckarter pushed to master at apple/swift
@jckarter
  • @jckarter ddc05a3
    Remove unused SILParameterInfo/SILResultInfo::subst helpers.
jckarter pushed to master at apple/swift
@jckarter
  • @jckarter 54903d2
    Remove unused PolymorphicFunctionType::substGenericArgs.
jckarter commented on pull request apple/swift#1157
@jckarter

@tienex For reference, a better solution than using -Bsymbolic long term might be to have swiftc emit public ELF symbols with protected visibility …

jckarter commented on pull request apple/swift#1194
@jckarter

@aschwaighofer and/or @nadavrot will probably want to look at this. I know we've gone back and forth on some of these micro optimizations as the hi…

jckarter commented on pull request apple/swift#1180
@jckarter

I agree that making the terminator explicit is independently a good idea. A better way to stage in the change might be to add an emitStringLiteralW…

jckarter commented on pull request apple/swift#1180
@jckarter

I see, I didn't realize it was handled by a builtin rather than a proper SIL instruction. I was saying that it might make sense for a dedicated SIL…

jckarter commented on pull request apple/swift#1180
@jckarter

Alternately, if this is only for code coverage, and the string literals belong in a separate section anyway, should we make the coverage-related in…

jckarter commented on pull request apple/swift#1180
@jckarter

It looks like LLVM's already smart enough to put non-null-terminated strings in __TEXT,__const instead of __cstring.

jckarter commented on pull request apple/swift#1180
@jckarter

The encoding shouldn't need to know whether to add a null or not, we can add the null to the string_literal value.

jckarter commented on commit apple/swift@4a6046e8c3
@jckarter

Not quite, you also have to shave the () off of deinit. The DI rules inside initializers will also probably trip you up if you accidentally tried t…

jckarter commented on pull request apple/swift#1180
@jckarter

I don't like this. I'd prefer to have it so that string_literal instructions explicitly include the null terminator when it's wanted. You should al…

jckarter commented on commit apple/swift@4a6046e8c3
@jckarter

I wonder if Bjarne's ever made this mistake.

@jckarter master is now 8c57b12
jckarter commented on pull request apple/swift#1157
@jckarter

(We've ended up exercising Mach-O relocations that weren't implemented in MCJIT and needed to be implemented in the past. You could ping llvm-dev a…

jckarter commented on pull request apple/swift#1157
@jckarter

Yeah, RuntimeDyld is part of LLVM's JIT though, it's not the real ld.so, so it'd be a bit easier for us to fix.

jckarter commented on pull request apple/swift#1157
@jckarter

Being illegal in shared objects shouldn't be a constraint on MCJIT. That's something LLVM ought to be able to fix.

jckarter commented on commit apple/swift@6a85b06129
@jckarter

@slavapestov Is this correct if the local var is lazy, or matches one of the other conditions that introduce accessors right after this check?

jckarter commented on pull request apple/swift#1157
@jckarter

@tienex Sure, but the behavior of -Bsymbolic is the right thing for Swift-generated symbols.

Something went wrong with that request. Please try again.