syd (Tom Sydney Kerckhove)
User

Projects

User does not belong to any projects.

User Details

User Since
Jul 18 2016, 4:29 AM (144 w, 2 d)

Recent Activity

Mar 4 2019

Marge Bot <ben+marge-bot@smart-cactus.org> committed rGHC2e9426df902c: hWaitForInput-accurate-socket test (authored by syd).
hWaitForInput-accurate-socket test
Mar 4 2019, 3:08 PM

Dec 10 2017

syd added a comment to D4103: hWaitForInput-accurate-pipe.

Looks like this one is done as well.

Dec 10 2017, 12:22 PM
syd added a comment to D4104: hWaitForInput-accurate-socket.

Looks ilke this one is done.

Dec 10 2017, 12:21 PM
syd updated the diff for D4095: hWaitForInput-accurate-stdin.

extra info

Dec 10 2017, 10:00 AM

Nov 26 2017

syd added a comment to D4012: base: fdReady(): Return only after sycall returns after `msecs` have passed.

I've had another sniff test, just to make sure that the +1 doesn't pile up for every rounding error and instead just fixes the rounding error as necessary, which seems to be the case.

Nov 26 2017, 11:46 PM

Nov 24 2017

syd added a comment to D4104: hWaitForInput-accurate-socket.

This passes on linux. I'll need harbormaster to verify that this works on windows as well.

Nov 24 2017, 10:22 AM
syd updated the diff for D4104: hWaitForInput-accurate-socket.

Fixed > -> >= again.

Nov 24 2017, 10:21 AM
syd updated the diff for D4104: hWaitForInput-accurate-socket.

Rebase onto master

Nov 24 2017, 10:21 AM
syd updated the diff for D4103: hWaitForInput-accurate-pipe.

Rebased onto master

Nov 24 2017, 10:02 AM
syd planned changes to D4096: Regression test for waiting too short a time.

This fails unexpectedly, at the moment. Will continue when it doesn't anymore.

Nov 24 2017, 9:56 AM
syd updated the diff for D4096: Regression test for waiting too short a time.

Put back a test that I accidentally deleted.

Nov 24 2017, 9:56 AM
syd updated the diff for D4096: Regression test for waiting too short a time.

Fixed the millisecond problem.

Nov 24 2017, 9:55 AM
syd updated the diff for D4095: hWaitForInput-accurate-stdin.

Fixed the millisecond problem.

Nov 24 2017, 9:12 AM
syd reclaimed D4096: Regression test for waiting too short a time.
Nov 24 2017, 9:12 AM
syd abandoned D4096: Regression test for waiting too short a time.
Nov 24 2017, 9:11 AM
syd updated the diff for D4096: Regression test for waiting too short a time.

Rebased on master

Nov 24 2017, 9:00 AM
syd updated the diff for D4095: hWaitForInput-accurate-stdin.

Fixed a > -> >=

Nov 24 2017, 8:21 AM
syd updated the diff for D4095: hWaitForInput-accurate-stdin.

Rebase ontop of master

Nov 24 2017, 7:46 AM

Nov 14 2017

syd added inline comments to D4011: base: fdReady(): Fix timeouts > ~49 days overflowing. Fixes #14262..
Nov 14 2017, 4:33 AM

Nov 8 2017

syd added a comment to D4127: Timeout benchmark: should be more, but not much more, than 100ms.

I'll admit I'm not really sure what to do here.
I'm quite certain that we want to avoid flaky tests.
We have been there before and it is terrible: a single flaky test severely compromises the usefulness of CI.

Nov 8 2017, 3:01 AM

Oct 30 2017

syd added a comment to D4127: Timeout benchmark: should be more, but not much more, than 100ms.

@Phyx, I used to have that test, see D4120.

Oct 30 2017, 1:26 PM

Oct 27 2017

syd added a comment to D4127: Timeout benchmark: should be more, but not much more, than 100ms.

I don't know either. It must be a benchmark, and not a test, but I'm not sure where to put "should be close to Xms" benchmarks that are not "as fast as possible" benchmarks.

Oct 27 2017, 9:11 AM

Oct 26 2017

syd created D4127: Timeout benchmark: should be more, but not much more, than 100ms.
Oct 26 2017, 5:01 AM
syd planned changes to D4096: Regression test for waiting too short a time.
Oct 26 2017, 4:24 AM

Oct 24 2017

syd added a comment to D4120: Remove the 'legroom' part of the timeout-accurate-pure test because of the flakyness it causes..

Let's remove the test entirely. I'll be making a benchmark soon, that benchmarks both of these conditions since they cannot actually be tests.

Oct 24 2017, 7:10 AM
syd updated the diff for D4120: Remove the 'legroom' part of the timeout-accurate-pure test because of the flakyness it causes..

Diffed off the wrong commit.

Oct 24 2017, 7:10 AM
syd updated the diff for D4120: Remove the 'legroom' part of the timeout-accurate-pure test because of the flakyness it causes..

Remove the test entirely

Oct 24 2017, 7:10 AM

Oct 23 2017

syd added a comment to D4120: Remove the 'legroom' part of the timeout-accurate-pure test because of the flakyness it causes..

Actually, nh2 has just brought it to my attention that even delta > 0 cannot be tested reliably. It's best if he explains it himself, but that means we better just remove this test entirely (and use the ones I am making for hWaitForInput instead).

Oct 23 2017, 5:09 PM
syd created D4120: Remove the 'legroom' part of the timeout-accurate-pure test because of the flakyness it causes..
Oct 23 2017, 7:48 AM
syd planned changes to D4103: hWaitForInput-accurate-pipe.
Oct 23 2017, 7:09 AM
syd planned changes to D4095: hWaitForInput-accurate-stdin.
Oct 23 2017, 7:09 AM
syd planned changes to D4104: hWaitForInput-accurate-socket.
Oct 23 2017, 7:09 AM
syd added a comment to rGHC13758c6cfec1: Added a test for 'timeout' to be accurate..

In fairness the test does seem to pass more often than not on OSX, but always fails on Windows.

OSX:
Failed https://phabricator.haskell.org/harbormaster/build/36469/

Oct 23 2017, 5:24 AM

Oct 22 2017

syd added a comment to rGHC13758c6cfec1: Added a test for 'timeout' to be accurate..

That's not the right way to do it. You have to mark the test as expect_fail and remove the expect_fail when the implementation is fixed.
Right now each and every build fails so we get emails for broken builds that are not our fault.

Oct 22 2017, 6:03 PM
syd added a comment to rGHC13758c6cfec1: Added a test for 'timeout' to be accurate..

Is this test supposed to run on all platforms? It's currently only passing on Linux and failing on everything else. maybe it should be marked as Linux only?

Oct 22 2017, 11:14 AM

Oct 18 2017

syd updated the diff for D4095: hWaitForInput-accurate-stdin.

Comment about FILE_TYPE_CHAR

Oct 18 2017, 11:10 AM
syd added a comment to D4104: hWaitForInput-accurate-socket.

Make sure to have an extra close look at this, I'm not so sure this will turn out to be a FILE_TYPE_SOCKET when it gets down to the RTS.

Oct 18 2017, 12:28 AM
syd created D4104: hWaitForInput-accurate-socket.
Oct 18 2017, 12:03 AM
syd added a parent revision for D4104: hWaitForInput-accurate-socket: D4103: hWaitForInput-accurate-pipe.
Oct 18 2017, 12:03 AM
syd added a child revision for D4103: hWaitForInput-accurate-pipe: D4104: hWaitForInput-accurate-socket.
Oct 18 2017, 12:03 AM

Oct 17 2017

syd updated the diff for D4074: Added a test for 'timeout' to be accurate..
  • Use monotonic time
Oct 17 2017, 11:01 PM
syd updated the diff for D4096: Regression test for waiting too short a time.

Made the legroom 1ms instead of 1s

Oct 17 2017, 11:01 PM
syd updated the diff for D4095: hWaitForInput-accurate-stdin.

Made the legroom 1ms instead of 1s

Oct 17 2017, 11:01 PM
syd updated the diff for D4103: hWaitForInput-accurate-pipe.

woops

Oct 17 2017, 11:01 PM
syd updated the diff for D4103: hWaitForInput-accurate-pipe.

Made the legroom 1 ms instead of 1s

Oct 17 2017, 11:01 PM
syd added a parent revision for D4103: hWaitForInput-accurate-pipe: D4096: Regression test for waiting too short a time.
Oct 17 2017, 11:01 PM
syd created D4103: hWaitForInput-accurate-pipe.
Oct 17 2017, 11:01 PM
syd added a child revision for D4096: Regression test for waiting too short a time: D4103: hWaitForInput-accurate-pipe.
Oct 17 2017, 11:01 PM
syd added a comment to D4095: hWaitForInput-accurate-stdin.

I'm quite unsure why this fails. It doesn't fail locally.

Oct 17 2017, 9:28 AM
syd added a comment to D4074: Added a test for 'timeout' to be accurate..

Thanks for the review. This looks good to me to, but for some reason the build fails. This diff used to say print instead of putStrLn, but the build still fails as if it said print. Is there anything I can do about that?

Oct 17 2017, 8:21 AM

Oct 16 2017

syd added a comment to D4096: Regression test for waiting too short a time.

Be aware that @nh2 has been looking at similar things in this area. See D4012 and related differentials.

Oct 16 2017, 10:12 PM
syd added a parent revision for D4096: Regression test for waiting too short a time: D4095: hWaitForInput-accurate-stdin.
Oct 16 2017, 11:32 AM
Herald added a reviewer for D4096: Regression test for waiting too short a time: austin.
Oct 16 2017, 11:32 AM
syd added a child revision for D4095: hWaitForInput-accurate-stdin: D4096: Regression test for waiting too short a time.
Oct 16 2017, 11:32 AM
syd added a child revision for D4074: Added a test for 'timeout' to be accurate.: D4095: hWaitForInput-accurate-stdin.
Oct 16 2017, 11:31 AM
syd added a parent revision for D4095: hWaitForInput-accurate-stdin: D4074: Added a test for 'timeout' to be accurate..
Oct 16 2017, 11:31 AM
Herald added a reviewer for D4095: hWaitForInput-accurate-stdin: austin.
Oct 16 2017, 11:31 AM
syd updated the diff for D4074: Added a test for 'timeout' to be accurate..
  • Use monotonic time
Oct 16 2017, 11:02 AM
syd updated the diff for D4074: Added a test for 'timeout' to be accurate..
  • Use monotonic time
Oct 16 2017, 10:14 AM
syd added a parent revision for D4074: Added a test for 'timeout' to be accurate.: D4079: Expose monotonic time from GHC.Event.Clock.
Oct 16 2017, 10:02 AM
syd added a child revision for D4079: Expose monotonic time from GHC.Event.Clock: D4074: Added a test for 'timeout' to be accurate..
Oct 16 2017, 10:02 AM
syd updated the diff for D4074: Added a test for 'timeout' to be accurate..
  • Use monotonic time
Oct 16 2017, 9:59 AM

Oct 9 2017

syd updated the diff for D4079: Expose monotonic time from GHC.Event.Clock.

Moved GHC.Event.Clock to GHC.Clock and exposed it

Oct 9 2017, 6:43 AM

Oct 8 2017

syd updated the diff for D4079: Expose monotonic time from GHC.Event.Clock.

Updating D4079: Expose monotonic time from GHC.Event.Clock

Oct 8 2017, 10:37 PM
syd planned changes to D4074: Added a test for 'timeout' to be accurate..

Will use monotonic time when D4079 lands.

Oct 8 2017, 6:18 AM
syd created D4079: Expose monotonic time from GHC.Event.Clock.
Oct 8 2017, 6:18 AM

Oct 6 2017

Herald added a reviewer for D4074: Added a test for 'timeout' to be accurate.: austin.
Oct 6 2017, 4:15 AM

Sep 29 2017

syd added inline comments to D4041: fdReady: Use C99 bools / CBool in signature.
Sep 29 2017, 8:46 AM

Jul 22 2016

syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • last last fixes?
Jul 22 2016, 10:47 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

Next round done.

Jul 22 2016, 10:47 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

This is getting slightly rediculous :D

Jul 22 2016, 4:39 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • last fixes?
Jul 22 2016, 4:38 AM

Jul 21 2016

syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Latest fixes
Jul 21 2016, 9:10 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

Latest fixes also done.

Jul 21 2016, 9:09 AM

Jul 20 2016

syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

These comments have been great.
Keep them coming :D

Jul 20 2016, 9:00 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • First round of comments.
Jul 20 2016, 9:00 AM

Jul 19 2016

syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Expansions in basicTypes
Jul 19 2016, 9:21 AM
syd added inline comments to D2406: Expanded one abbreviation in documentation #12405.
Jul 19 2016, 7:28 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Fixed typos
Jul 19 2016, 7:27 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • More expansions and a correction
Jul 19 2016, 7:26 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • More expansions in HsSyn.hs and HsType.hs
Jul 19 2016, 3:56 AM

Jul 18 2016

syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Fixed a few documentation compilation errors
Jul 18 2016, 4:17 PM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

Even more expansions.

Jul 18 2016, 10:53 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Even more expansions!
Jul 18 2016, 10:52 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Expansions in hsSyn/HsDoc.hs and hsSyn/HsExpr.hs
Jul 18 2016, 10:29 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

HsBinds.hs is almost done, There's just one thing I did not understand. What is an ABExport?

Jul 18 2016, 9:26 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Expansions in hsSyn/HsBinds.hs
Jul 18 2016, 9:25 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

For review: Please check that the expansions are in fact correct.
I am not sure whether Doc should be Document or Documentation and I did not know what Bndr was so I assumed it was Binder.

Jul 18 2016, 8:13 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

Voila, all expansions in this file for top level types are done.
Would you like me to include expansions for other files in this diff as well?

Jul 18 2016, 8:11 AM
syd updated the diff for D2406: Expanded one abbreviation in documentation #12405.
  • Expanded abbreviations in HsDecls
Jul 18 2016, 8:10 AM
syd added a comment to D2406: Expanded one abbreviation in documentation #12405.

Thank you for the abbreviations page!

Jul 18 2016, 7:20 AM
syd retitled D2406: Expanded one abbreviation in documentation #12405 from to Expanded one abbreviation in documentation #12405.
Jul 18 2016, 4:32 AM