IRGen: Actually make use of private linkage #15476
Conversation
|
@swift-ci Please test Linux |
|
Build failed |
slavapestov
force-pushed
the
private-linkage
branch
from
fdc0ac6 to
4d7fed1
Compare
|
@swift-ci Please test |
|
@swift-ci Please test source compatibility |
|
Build failed |
|
@swift-ci Please test |
|
@swift-ci Please test source compatibility |
|
Build failed |
There was a problem hiding this comment.
Nice cleanup.
The HiddenNonUnique change should be NFC since we never actually assign that linkage.
The internal linkage change LGTM assuming the debugger does not need access to private symbols (I don't know the history there). @jrose-apple added this code in 2015.
|
Yeah, the lldb tests passed so I’m assuming that hack is no longer necessary. But I’ll wait for @rjmccall or @jrose-apple to take a look before merging. I also noticed we emit value witnesses as linkonce_odr for no apparent reason. I don’t think they need to be ODR except for imported types. |
include/swift/SIL/FormalLinkage.h
Outdated
| /// This entity is visible in only a single Swift module but does not | ||
| /// have a unique file that is known to define it. | ||
| HiddenNonUnique, | ||
| Hidden, |
There was a problem hiding this comment.
Removing HiddenNonUnique makes sense to me if we never use it currently. Renaming HiddenUnique seems like it doesn't actually win us anything, and it just creates pain if we ever need to reintroduce HiddenNonUnique.
|
I'm not quite sure what past me was thinking. LLVM |
|
@slavapestov nice! |
slavapestov
force-pushed
the
private-linkage
branch
from
4d7fed1 to
98d74bd
Compare
|
@swift-ci Please smoke test |
No description provided.