Thanks for looking into this. I've reverted the test assertions on the
branch in r352354.
On Mon, Jan 28, 2019 at 1:14 AM Sam McCall wrote:
>
> This is kind of expected: r347205 was a bugfix.
> Reverting it causes a real bug, and the tests are failing. The bug is
> acceptable for 8.0 - it affect
This is kind of expected: r347205 was a bugfix.
Reverting it causes a real bug, and the tests are failing. The bug is
acceptable for 8.0 - it affects only a marginal case (go-to-definition on a
#include *outside* the preamble section of a file that was also
transitively #included in the preamble).
Sorry to bring more bad news, but merging this seems to have regressed
a clangd test on the branch (I didn't notice at the time, because I
ran the wrong set of tests, d'oh).
I've searched my inbox but couldn't find any recent commits touching
the test. Do you have any idea what might be happening?
Merged to 8.0 in r352225.
On Thu, Jan 24, 2019 at 10:55 AM Sam McCall via cfe-commits
wrote:
>
> Author: sammccall
> Date: Thu Jan 24 10:55:24 2019
> New Revision: 352079
>
> URL: http://llvm.org/viewvc/llvm-project?rev=352079&view=rev
> Log:
> [FileManager] Revert r347205 to avoid PCH file-descr
Author: sammccall
Date: Thu Jan 24 10:55:24 2019
New Revision: 352079
URL: http://llvm.org/viewvc/llvm-project?rev=352079&view=rev
Log:
[FileManager] Revert r347205 to avoid PCH file-descriptor leak.
Summary:
r347205 fixed a bug in FileManager: first calling
getFile(shouldOpen=false) and then get
