Re: [Boost-build] Adding command line arguments to tests

Subject: Re: [Boost-build] Adding command line arguments to tests
From: Johan Nilsson (r.johan.nilsson_at_[hidden])
Date: 2009-11-17 08:00:56

Christian Auby wrote:
> I'm running the build from Pulse continous integration server, and I
> want to capture the test results. Instead the build fails so it says
> "tests: none".

How do you capture the test results? IIRC Boost.Test has no built-in support 
for writing XML to file, and you are likely to have xml mixed with raw 
console output if you just tell Boost.Test to generate xml, and then run 
bjam. Can Pulse handle this? Also, can Pulse handle the generated Boost.Test 
XML format? It differs from the JUnit XML format.

I've in the past had to setup Boost.Test programmatically in order to write 
XML to file, but haven't really checked for built-in support for this during 
the last couple of years(!). Been using this for integration with 
CruiseControl and Hudson.

>
> Is there a way to make it continue on test errors, create the report,
> and only report target errors if a non-test target has a compilation
> error?

It (bjam) will continue on test errors unless you specify "-q" on the 
command-line. The final status will however be reported as failure if any 
target fails to build. As far as ignoring unit-test target failures - what 
if a test target has a _build_ error (compile/link etc)?

/ Johan