Re: [Boost-testing] [regression] Is there more information available?

Subject: Re: [Boost-testing] [regression] Is there more information available?
From: Steven Watanabe (watanabesj_at_[hidden])
Date: 2012-10-20 18:52:41

AMDG

On 10/20/2012 01:43 PM, Robert Ramey wrote:
> I'm find myself wanting to know about the regression testing process.  I've 
> been looking at ......\tools\regression\src and have a few questions.  It 
> seems that most of my questions
> can only be answered by trolling through the python source.  However this is 
> pretty
> tedious and I'm wonder if there is more information.  The fact that I don't 
> know anything
> about python doesn't help either.
> 
> and here are some random questions.
> 
> a) There is a python script for process_jam_log.  How is this different from 
> the process_jam_log compiled executable?
> 

It's a port to Python of the C++ code.  It's not
generally used and is probably out of date by now.

> b) I gather that the bjam.log file is uploaded along with a bunch of 
> concatonated test_log.xml files.  Is there more information about this
> 

By default, I believe that run.py uploads a
single xml file produced by process_jam_log.

> c) It seems there is code for uploading to a "dart" server.  Is there any 
> more information about this?
> 

I have no idea.  I suspect this was someone's
attempt to make a new system which was never
finished.

> d) there is a directory xsl_reports / v2 (and others) with a bunch of xslt 
> scripts for producing html output from xml.  Presumably this xml is 
> generated somewhere.  It's not obvious to me how the data flows from 
> bjam.log and test_xml.log through to the generation of the test matrix on 
> the server.  Any insight appreciated.
> 

There's a script called build_results.sh which
pulls the logs from the ftp server, builds the
html using the XSL scripts and uploads the results
to the Boost website.  I believe that Noel runs
this for us.

In Christ,
Steven Watanabe