propagate TIME_DEPTH to the helper threads for -Z time-passes #38645
Conversation
|
@bors: r+ |
|
📌 Commit ad747c5 has been approved by |
|
@bors p=1 Time-sensitive, needed to fix perf results for incr. comp. tests. |
|
⌛ Testing commit ad747c5 with merge 26366b2... |
|
💔 Test failed - status-travis |
|
@bors: retry |
|
⌛ Testing commit ad747c5 with merge 8ad040c... |
|
💔 Test failed - status-travis |
| thread_local!(static TIME_DEPTH: Cell<usize> = Cell::new(0)); | ||
|
|
||
| /// Read the current depth of `time()` calls. This is used to | ||
| /// encourage indentation across threads. |
There was a problem hiding this comment.
This comment seems to be nonsensical given that TIME_DEPTH is a thread-local static. Perhaps it should be an AtomicUsize and shared across threads instead?
| /// `set_time_depth()` with the result from `time_depth()` in the | ||
| /// parent thread. | ||
| pub fn set_time_depth(depth: usize) { | ||
| TIME_DEPTH.with(|slot| slot.set(depth)); |
|
@bors: retry
* network error
…On Wed, Dec 28, 2016 at 12:19 PM, bors ***@***.***> wrote:
💔 Test failed - status-travis
<https://travis-ci.org/rust-lang/rust/builds/187295305>
—
You are receiving this because you commented.
Reply to this email directly, view it on GitHub
<#38645 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AAD95J-wlDYInax78fw3gD-042r6R_Rkks5rMsRMgaJpZM4LWoad>
.
|
|
⌛ Testing commit ad747c5 with merge 07fdfe6... |
|
💔 Test failed - status-travis |
|
@bors: retry
|
Yes. |
|
⌛ Testing commit ad747c5 with merge ae38ba2... |
|
💔 Test failed - status-appveyor |
propagate TIME_DEPTH to the helper threads for -Z time-passes Currently, the timing measurements for LLVM passes and the like don't come out indented, which messes up `perf.rust-lang.org`. r? @nrc
|
☀️ Test successful - status-appveyor, status-travis |
Currently, the timing measurements for LLVM passes and the like don't come out indented, which messes up
perf.rust-lang.org.r? @nrc