Subject: Re: [Boost-build] Toolset names and target alternatives
From: Gevorg Voskanyan (v_gevorg_at_[hidden])
Date: 2011-07-14 15:45:31
Edward Diener wrote:
> To: boost-build_at_[hidden]
> Sent: Wed, July 13, 2011 1:14:21 AM
> Subject: [Boost-build] Toolset names and target alternatives
> Currently I can specify a toolset name by its 'name' or its 'name-version'. But
>there is no way to specify a toolset 'name' so that all versions starting with
>some beginning part of the version number are specified.
> As an example I can say 'gcc' for a toolset name or 'gcc-4.5.2' for a toolset
>name. The first encompasses all gcc toolsets of whatever version number and the
>second encompasses specifically gcc version 4.5.2. But there is no way of
>specifying all gcc version 4 toolsets.
> There should be.
> This would be particular useful with target alterantives. As an example I want
>to setup a jam target alternative for all gcc 4 versions in order to test a
>feature of the -std=c++0x option. This option does not exist for gcc version 3
>releases but does exist for gcc version 4 releases. But to do this I have to
>repeat the alternative for each individual gcc 4 version. So if I have:
> test-suite sometest_gcc
> [ compile somesource.cpp
> : <cxxflags>-std=c++0x
> : somesource_gcc
> I must repeat this for each gcc 4 version. Of course in a typical case there
>will be many tests, so I end of repeating huge sets of tests for every specific
>gcc versions I want to setup with a taret alternative. Ugh !
> Please consider treating <toolset>gcc-4 as all versions which start with gcc-4
>such as gcc-4.3.0, gcc-4.4.0 etc. Also <toolset>gcc-4.5 should mean all version
>of gcc which start with gcc-4.5 such as gcc-4.5.0, gcc-4.5.2 etc.
> This ability to represent more than one version via a partial version
>identifier would be welcome in Boost Build.
I'm afraid that may not adequately solve your problem as -std=c++0x option is
only available from gcc 4.3 onwards.
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