Boost Testing :
Subject: Re: [Boost-testing] [EXTERNAL] Re: Help: my tester only does one test case!
From: Belcourt, Kenneth (kbelco_at_[hidden])
Date: 2014-03-13 00:14:22
On Mar 12, 2014, at 12:33 AM, Steve M. Robbins wrote:
> On Tue, Mar 11, 2014 at 04:29:46AM +0000, Belcourt, Kenneth wrote:
>> On Mar 10, 2014, at 10:24 PM, Steve M. Robbins wrote:
>> On March 10, 2014 07:20:09 PM Steven Watanabe wrote:
>> The test-type field appears to be broken. It's most
>> likely a problem with process_jam_log. (The Boost.Build
>> tests are run separately, and don't use process_jam_log)
>> So where is process_jam_log run: on my machine or on the one that gathers
>> uploaded results ? Why would only my machine be affected?
>> On your machine. Check that process_jam_log, the executable, can be found by run.py. If you capture the git checkout + bjam build, you should also see it build process_jam_log, unless it finds an existing process_jam_log. It may be trying to build it, and fail. You should see something like this in the log (this for master).
> Thanks, yes, that helps. So I do see the lines above where it starts
> to build process_jam_log. And later in the log it reports success:
> ... and that is the last mention of process_jam_log in the jenkins build log.
Correct. Do you have about 6500 lines of
# Processing test log "yada
after bjam finishes running the tests (this with develop)?
> So what are the "test-type field" values supposed to look like? I se
> a mix of "", "run" and "lib" in my results file.
Yep, this from a recent run of master.
-bash-4.1$ grep "test-type=\"\"" Sandia-Linux.xml | wc
43 344 8404
-bash-4.1$ grep "test-type=\"lib\"" Sandia-Linux.xml | wc
86 688 17184
-bash-4.1$ grep "test-type=\"run\"" Sandia-Linux.xml | wc
6155 49113 1769779