Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Nov 30 04:20
    billcxx closed #315
  • Nov 30 04:20
    billcxx commented #315
  • Nov 29 19:17
    Snaipe commented #315
  • Nov 29 19:06
    Snaipe commented #315
  • Nov 29 18:40
    billcxx opened #315
  • Nov 26 17:28
    Snaipe commented #314
  • Nov 22 17:11
    Bensuperpc edited #314
  • Nov 22 17:09
    Bensuperpc edited #314
  • Nov 22 17:08
    Bensuperpc opened #314
  • Nov 21 00:40
    karlvignon commented #223
  • Nov 13 21:29
    kugel- commented #172
  • Nov 13 20:49
    Snaipe commented #312
  • Nov 13 20:29
    Snaipe labeled #313
  • Nov 13 20:29
    Snaipe labeled #313
  • Nov 13 20:29
    Snaipe labeled #313
  • Nov 13 20:29
    Snaipe opened #313
  • Nov 13 20:26
    Snaipe commented #172
  • Nov 13 15:12
    kugel- commented #172
  • Nov 11 17:41
    jonathanturcotte commented #312
  • Nov 06 16:25
    jonathanturcotte commented #312
Franklin Mathieu
@Snaipe
It might be something I've forgotten
or simply something not included in the JUnit XML spec
okay, it was forgotten
there is definitely a "time" attribute
For both "testsuite" and "test" it seems
Dominik
@kaidowei
should I open an issue or create a pullrequest
or do you want to fix it
Franklin Mathieu
@Snaipe
I'm currently working on the fork patch for nanomsg, so go ahead if you want to fix it
just note that the spec says that the time is an integer representing the number of seconds
so you'll need to round to the nearest integer
Dominik
@kaidowei
which (in almost all cases) makes no sense :(
Franklin Mathieu
@Snaipe
yeah, it's a shame
I think the idea was to have more detail in the "timestamp" field
but currently "timestamp" is not an information that is gathered in the stats
actually nevermind, there's no "timestamp_end", so "timestamp" in itself is useless to calculate the time spent
and in any case a timestamp is in seconds
so yeah, it's too bad.
Dominik
@kaidowei
meaning 0.001 is 1ms
Franklin Mathieu
@Snaipe
Oh, then that's perfect
I don't know why I remember it being an integer
I must have mixed it up with something else
so then round up to the millisecond
in any case it's the only sane measurement accross platforms, as some does not give precision up to nanoseconds
Dominik
@kaidowei
elapsed_time is already in ms?
Franklin Mathieu
@Snaipe
seconds
Dominik
@kaidowei
ah so, I don't have to do anything, cool
Franklin Mathieu
@Snaipe
yep
Dominik
@kaidowei
what does it contain, if the test crashed?
Franklin Mathieu
@Snaipe
0
because nothing gets reported after the crash
I ought to fix this someday
Dominik
@kaidowei
the time for a suite... do I have to calc that or is there already a function somewhere?
Franklin Mathieu
@Snaipe
no, you have to sum all the elapsed_times
Dominik
@kaidowei
best place for that?
Franklin Mathieu
@Snaipe
Dominik
@kaidowei
yeah... are there unittests for the xml io?
Franklin Mathieu
@Snaipe
there are no unit tests, but a bunch of system tests to validate the output in test/cram
if you want to regenerate the new tests outputs, simply call make cram_tests CRAM=-i
Dominik
@kaidowei
pullrequest to master or bleeding?
Franklin Mathieu
@Snaipe
bleeding
I'll roll this out for 2.3.0
Dominik
@kaidowei
are you currently on bleeding?
can you check for me, if the time is accounted correctly?
I have a sleep in my test and the time is still 0.00
with 2.2.1 it works
Franklin Mathieu
@Snaipe
mmh, right. Times aren't reported on bleeding for some reason
let me investigate
oh, right, I remember why
that's because the protocol started working with timestamps rather than elapsed times