From: Markus Schöpflin (markus.schoepflin_at_[hidden])
Date: 2006-02-17 05:59:29
Vladimir Prus wrote:
> On Tuesday 14 February 2006 18:00, Markus Schöpflin wrote:
>> Vladimir Prus wrote:
>>>> Is there a way to achieve this without messing around with a file living
>>>> _inside_ the boost source tree?
>>> Yes, create user-config.jam in your home directory (the value of $HOME
>>> environment variable), and make all customization there.
>> Yes, that works.
>> It seems that BOOST_BUILD_PATH is not searched for this file, is there any
>> specific reason for this?
> It fact, it should search BOOST_BUILD_PATH. That how it find the one inside
> Boost.Build tree in case where there's no in $HOME.
I think this is broken, then.
> echo $BOOST_BUILD_PATH
> ls /vol2/boost/user-config.jam
> bjam -d+2 -n --v2 gcc-4_0_2_tru64 debug
warning: Python location is not configured
warning: the Boost.Python library won't be built
Building Boost.Regex with the optional Unicode/ICU support disabled.
Please refer to the Boost.Regex documentation for more information
(and if you don't know what ICU is then you probably don't need it).
warning: no toolsets are configured.
warning: you won't be able to build C++ programs.
warning: please consult the documentation.
notice: could not find main target gcc-4_0_2_tru64
notice: assuming it's a name of file to create
error: unable to construct ../../../libs/filesystem/build/boost_filesystem
To debug this problem, try the --debug-generators option.
It works if I move user-config.jam to my home directory.
>>> The previous code put options to the second argument -- which is the
>>> 'command' argument.
>> Now it's working, thank you very much for your patience.
>> So the procedure most regression runners will have to follow is:
>> 1. Create a user-config.jam in their home directory, or wherever this file
>> is allowed to live.
>> 2. Add customized toolsets to this file.
>> 3. Use <toolset>-<version1>, <toolset>-<version2>, ... instead of using
>> <customized toolset 1>, <customized toolset 2>, ... when running the
>> regression tests.
> Yes, and I've tried to document that on:
Good, this will help other regression runners, I hope.
>> I just started a boost regression run using bbv2, it will take some days to
>> complete. When it's finished I will do another run using bbv1 and then try
>> to compare the results. We'll see how this fares.
> Will be interesting to see.
Phew, the V2 run finally completed and post processing the log file when
well. But now an ugly problem is rearing its head again when trying to
create the report.
> ./dist/bin/compiler_status --v2 /vol2/boost/src/boost-HEAD boost_v2.html
**** exception(205): std::runtime_error:
******** errors detected; see standard output for details ********
This is an error from boost.filesystem which has been there since the
beginning of boost.filesystem, IIRC, but so far always only showed up in
the regression tests and never in real live scenarios. All boost.filesystem
test errors can be attributed to this problem, AFAICT.
I once tried to debug it but gave up after several hours, because I
couldn't find anything wrong with the code. It looked like readdir_r being
broken on this platform, but when trying to reproduce the error with
readdir_r alone, I never succeeded.
Oh well, I will try once more to hunt down this bug, maybe I will have more
luck this time.
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk