Boost Testing :
From: Rene Rivera (grafik.list_at_[hidden])
Date: 2005-01-06 00:07:07
Sorry Stephen... I'm afraid it's going to take a bit of effort to get
CodeWarrior on MacOS working...
Stephen W. Carson wrote:
> My first attempt to run the regressions against CodeWarrior on the
> Mac didn't get very far.
> I ran "python regression.py --runner=Dirxion --toolsets=cw" and,
> eventually, got "Unknown toolset" from build.sh when it tried to build
> bjam even though "cw" is on the list of supported toolsets:
That's the list of toolsets that Boost.Build supports. As it turns out
that's a different set than the toolsets that Boost.Jam supports
> (build.sh says that it can only handle "acc, como, darwin, gcc,
> intel-linux, kcc, kylix, mipspro, sunpro, tru64cxx, vacpp" which is a
> rather smaller list than the toolsets that the web page lists as
> supported. Why do these lists differ?!?)
Different build systems if you can believe it :-) The build.sh/bat are
the bare bones boot strapping to get bjam compiling. The list of
toolsets that the build.sh/bat scripts support is..
http://www.boost.org/tools/build/jam_src/index.html#building_bjam - As
you can see it's a rather limited list. :-( But there's an immediate way
around this. You can specify which toolset to use to build bjam with to
the regression.py script with the "--bjam-toolset" option. So if you
have gcc available you can use it to build bjam and get farther along.
The other option would be hacking at build.sh and build.jam in
boost-root/tools/build/jam_src to add the support for CW in MacOS.
But it doesn't end there. Even though the Getting Started docs say that
CW is supported, it currently only supports it on Windows. It is an
often requested addition though but no one has had the time to add
support for it.
> To build my confidence I went ahead and did a darwin run and you can
> see those results on the web site today. (Also, I've already updated my
> comments.html to be more informative).
You should also probably mention what version of MacOS.
> One more question, John Maddock suggested that I give the different
> toolset runs a "different identifier name"? What does he mean? The way
> the tables look it seems like I should use "--runner=Dirxion" for both
> darwin and CodeWarrior and the results will be differentiated by the
> differing toolset identifiers. Is there something else I should do to
> differentiate the results from the two compilers?
He means to make a custom toolset that includes the version information.
For example when I've run the Windows-CW-83 tests in the past I added
this file to my boost-root/tools/build/v1 directory:
#~ Copyright 2004 Rene Rivera.
#~ Distributed under the Boost Software License, Version 1.0.
#~ (See accompanying file LICENSE_1_0.txt or
# The following #// line will be used by the regression test table
# program as the column heading for HTML tables. Must not include
local CW_VERSION = 8.0 ;
extends-toolset cw ;
To indicate the CW version and the fact that I run the tests in release
mode instead of the regular debug mode.
Sorry about the wrapping.. you get the idea ;-) And there's other such
toolsets in that v1 dir as additional examples.
You should also probably mention what version of MacOS
-- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com - 102708583/icq