Boost logo

Boost-Build :

Subject: Re: [Boost-build] Issue in glob rule when invoked in <conditional>.
From: Steven Watanabe (watanabesj_at_[hidden])
Date: 2012-07-24 10:15:13


AMDG

On 07/24/2012 04:09 AM, Vladimir Prus wrote:
> On 24.07.2012 14:46, Jurko Gospodnetić wrote:
>> Hi.
>>
>>>>> indirect_conditional.py updated and is now failing with the trunk.
>>>>
>>>> FAOD, want me to commit the patch?
>>>
>>> I'm playing around with it just now, but don't hold back on my
>>> account - seems to work and is easy enough to revert if needed.
>>> :-)
>>
>> For what it's worth - Boost Build tests pass with msvc & Cygwin/gcc
>> & the Boost library 1.50 release builds fine with the patch.
>>
>> Patch seems fine, other than making me realize that I do not
>> completely understand or feel familiar with different global contexts
>> that
>> get used in Boost Build, e.g. 'current project' or 'current target'. :-)
>
> Well, to be honest I am not familiar either ;-) project.current is
> something that used to be set when reading project, so that helper
> rules defined in Jamfile declare metatargets in right project when
> children Jamfile call it.
>
> There was no way to get 'current metatarget' or 'current project' when
> generating a metatarget. So, my patch in fact introduces such
> context -- and I have not considered any longer-term strategy.
>

This makes me pretty nervous. I really
don't like hacking in features as someone
requests them. I'm not even convinced
that this is the correct behavior.
If we're going to do something like
this the documentation of glob needs
to be updated to make it clear that
it searches the 'current project' instead
of the project of the current module.
Also, how would the other project rules
need to be updated to be consistent?

In Christ,
Steven Watanabe


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