Times reported in XML for testcases are wrong

[avdv]

I am using specs2 version 4.4.1.

When running with showtimes the printed times are OK:

[info] ConvertControllerSpec
[info] ConvertController should
[info]   + convert a given Story to JSON API (340 ms)
[info]   + convert a given Story with story content to JSON API (249 ms)
[info]   + convert a given Story with several tags to JSON API (250 ms)
[info]   + convert an Image-Gallery Asset to JSON API (203 ms)
[info]   + convert a given Video Channel to JSON API (209 ms)
...
[info] Total for specification ConvertControllerSpec
[info] Finished in 12 seconds, 154 ms
[info] 47 examples, 0 failure, 0 error

But looking at the test-report in target/test-reports/controllers.ConvertControllerSpec.xml the times are totally off:

<testcase name="ConvertController should::convert a given Story to JSON API" classname="controllers.ConvertControllerSpec"
   time="8.757" />
<testcase name="ConvertController should::convert a given Story with story content to JSON API" classname="controllers.ConvertControllerSpec"
   time="8.401" />
<testcase name="ConvertController should::convert a given Story with several tags to JSON API" classname="controllers.ConvertControllerSpec"
   time="8.138" />
<testcase name="ConvertController should::convert an Image-Gallery Asset to JSON API" classname="controllers.ConvertControllerSpec"
   time="6.960" />
<testcase name="ConvertController should::convert a given Video Channel to JSON API" classname="controllers.ConvertControllerSpec"
   time="6.299" />

In reference to #613, I am not using sequential.

[avdv]

I had a quick look, attaching a debugger. The problem seems to be that the timers at https://github.com/etorreborre/specs2/blob/master/junit/shared/src/main/scala/org/specs2/reporter/JUnitXmlPrinter.scala#L45 are still running (timer.isStarted is true).

[etorreborre]

Thanks for having a look. The way the times were taken was quite wrong. I have a fix for it, wait a bit and a new version will be published later on today.

[avdv]

Awesome! Thank you Eric. 💯

[avdv]

BTW, while you're at it, I noticed the scaladoc for HmsTimer.stop is wrong too. It does not return a String at all.

[etorreborre]

Thanks I will change that too (and put a return type, I didn't use to do that when I started Scala!). Also you might have to be a bit patient. There's a failing spec that I need to investigate after this change.

[avdv]

Also you might have to be a bit patient. There's a failing spec that I need to investigate after this change.

No problem, it's not a big deal really. I just recently noticed CircleCI reported the slowest test alone took 33 seconds which piqued my interest. Historically, the test reports on our old Jenkins instance also reported such long runtimes, but nobody noticed until now.

[avdv]

FYI, I did just try version 4.5.0; looks good to me. 👍

[etorreborre]

Thanks for the feedback. I'm going to publish a 4.5.1 though because I think I broke a piece of some exotic functionality (random execution of examples).

[avdv]

OK, I'm holding my breath until the official announcement anyway... 😉

[etorreborre]

OK, just announced it: https://twitter.com/specs2org/status/1105071256112631809

[avdv]

Thank you! 💐