Boost logo

Boost-Build :

From: Rene Rivera (grafik666_at_[hidden])
Date: 2002-06-14 10:24:03


On 2002-06-14 at 06:59 AM, david.abrahams_at_[hidden] (David Abrahams) wrote:

>
>From: "David Abrahams" <david.abrahams_at_[hidden]>
>
>> ----- Original Message -----
>> From: "David Abrahams" <david.abrahams_at_[hidden]>
>>
>> > > Oh... that would only require that target names do not clash with
>> toolset
>> > and
>> > > build variants
>> >
>> > ...or that there's a way of disambiguating...
>>
>> namely, explicit specification of the property name for implicit features
>> which conflict with target names: <toolset>borland
>
>Y'know,
>
>Putting gristed names in the command-line really sucks. I can't count the
>number of times I've written
>
> -sBUILD=<cxxflags>blah
>
>and unintentionally created a file called "cxxflags"
>
>I mean, really: shouldn't we be using something like
>
> bjam gcc stlport-4.5.3 --inlining=off ...
>
>In which case for the non-implicit features we should be internally
>translating --feature=value into <feature>value. My only real question is
>whether we need an additional level of namespacing for feature settings.
>Something like --with-, e.g.
>
> bjam gcc --with-inlining=off
>
>Just to keep them out of the way of "global" command-line options.

That sounds nice... but if we are going the "configure" like route we might
as well map to it as closely as possible. Which means that it would be:

"bjam gcc --enable-inlining", or "bjam gcc --enable-inlining=yes"
and
"bjam gcc --disable-inlining", or "bjam gcc --enable-inlining=no"

The "with" type options are used for package options. For us that could
enable us to do:

bjam gcc --with-stlport

-- grafik - Don't Assume Anything
-- rrivera_at_[hidden] - grafik_at_[hidden]
-- 102708583_at_icq - Grafik666_at_AIM - Grafik_at_[hidden]

 


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