Page MenuHomePhabricator

freik (Kevin Frei)
Engineering Manager (Facebook)

Projects

User does not belong to any projects.

User Details

User Since
Nov 5 2014, 11:06 AM (250 w, 4 d)

Recent Activity

Apr 4 2016

freik updated subscribers of D18762: Rewrite Aggressive Dead Code Elimination.
Apr 4 2016, 9:54 AM

Feb 7 2015

freik requested changes to D7483: Terrible change as an example for CGO!.

Thanks for doing the LLVM tutorial, but this code clearly needs more work :)

Feb 7 2015, 9:56 AM

Nov 10 2014

freik added a comment to D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC..

Yup. Thanks!

Nov 10 2014, 7:33 AM

Nov 7 2014

freik updated the diff for D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC..

[Responding to feedback]
Updated the test case to be better directed (specifically testing externally_initialized and appending attributes). Moved the fix to be in GlobalVariable::copyAttributesFrom instead of at the copy site.

Nov 7 2014, 4:05 PM

Nov 6 2014

freik added a comment to D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC..

Fair enough. I'll go ahead & update the diff with the fix moved into copyAttributesFrom, and fix up handful of missing attributes in GlobalObject and GlobalValue, too.

Nov 6 2014, 12:44 PM
freik added a comment to D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC..

The reason I wasn't sure which was correct is because isExternallyIntialized is a default ctor parameter, so if someone's specifying a different value, then calling the copyAttributesFrom to fill in everything else, it's a bug. From a design perspective, having an API like copyAttributesFrom for a type with a ctor that has default parameters seems incredibly confusing. Perhaps I should eliminate the default parameter? And maybe this discussion should move over to llvmdev?

Nov 6 2014, 9:22 AM

Nov 5 2014

freik added a comment to D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC..

@rnk: I considered both, and am new to the codebase, so the 'correct' place to fix this wasn't apparent. Does anyone else have an opinion, here? Either way is effective, and pretty straightforward as well.

Nov 5 2014, 8:42 PM
freik updated the diff for D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC..

Missed the new test file...

Nov 5 2014, 3:56 PM
freik retitled D6145: Fix for nasty LTO issue that's most obvious in ObjectiveC. from to Fix for nasty LTO issue that's most obvious in ObjectiveC..
Nov 5 2014, 3:52 PM