Boost logo

Boost Testing :

From: Rene Rivera (grafikrobot_at_[hidden])
Date: 2007-08-28 00:34:25

Robert Ramey wrote:
> Boris Gubenko wrote:
>> Robert Ramey wrote:
>>>> From the --help I don't see a way to do this
>>> with having my system either download a tarball or update
>>> from SVN. What I want to do is just run the regression
>>> test on my system with my current setup and without forwarding
>>> the results (I assume that this latter is obtained with
>>> --dont-send-bjam-log).
>> You, probably, meant: "withOUT having my system either ..."
>> Here is how you can run tests in separate steps:
>> python get-source ... # downloads tarball
>> python setup ... # builds bjam, process_jam_log,
>> ...etc python test ... # runs tests
>> python collect-logs ... # creates XML log
>> python upload-logs ... # uploads logs
>> "..." is switches you usually specify on command line.
>> Different commands above require different switches (for example, you
>> don't need --runner on get-source command), but it is not an error to
>> specify them all on every command.
>> I guess, what you need is:
>> python test ...
>> python collect-logs ...
> I looked at this and its not clear to me that its going to test what I
> wanted
> to test.

Those two would run all the tests and run process_jam_log only. As
opposed to trying to build the various tools, and uploading the
<runner>.xml file to meta-comm.

> But when I look
> at collect logs it seems like it doesn't actually parse the test_log.xml
> but just sends it somewhere.

Hm, well it effectively concatenates all the test_log.xml files into a
single <runner>.xml file. It also happens to create a separate dart.xml
file which it send to the Dart2 server.

> Note that I don't actually know
> python but the 1019 script looks sort of C/shell script like and
> of course it does have 23 lines of helpful comments.

Yea, don't I know it :-(

> Sooo
> its not clear if/how I can verify that the xml file generated
> by process_jam_log won't create havoc downstream.

I don't think any one but meta-comm can tell you just by looking if the
results file is OK. So the only way to be sure is to try and submit some

> There
> is no separately documented format requirements for these
> files.

There is a tools/regression/xsl_reports/test_results.xsd which could be
used to do some rudimentary validation of the content.

...Perhaps looking at how I do the incremental testing on my Linux
machine can tell you how to run things on a real boost trunk checkout

Other than that if you can manage to run the "test" and "collect-logs",
you can then just upload the <runner>.xml file to the ftp server manually.

-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. -
-- rrivera/ - grafik/
-- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo

Boost-testing list run by mbergal at