[Boost-testing] --pjl-toolset option broken in run.py?

Subject: [Boost-testing] --pjl-toolset option broken in run.py?
From: Stuart A. Malone (samalone_at_[hidden])
Date: 2009-01-15 10:24:49

Hi folks,

I've been trying to help John Maddock test TR1 on Mac OS X, and in the  
process I thought I'd try running the full set of regression tests on  
a spare Mac that we keep for testing.

I ran into a problem with the run.py script trying to build  
process_jam_log.  By default, it tries to build process_jam_log using  
the gcc toolset rather than the darwin toolset.  The build then fails.

After doing some more reading, I discovered the --pjl-toolset flag to  
run.py, and added --pjl-toolset=darwin to my command line.  However,  
that flag seems to be broken.  Rather than adding "toolset=darwin" to  
the bjam command line for process_jam_log, it only added "darwin".   
The result was that process_jam_log still got built with the wrong  
toolset, and bjam complained that it couldn't find the target "darwin".

Am I correct that run.py isn't handling the --pjl-toolset flag  
correctly?

--Stuart