Re: [boost] Boost.Threads, the Boost.Build system and GCC

From: David Abrahams (abrahams_at_[hidden])
Date: 2001-06-09 07:17:08

----- Original Message -----
From: "John Maddock" <John_Maddock_at_[hidden]>
To: <boost_at_[hidden]>
Sent: Saturday, June 09, 2001 7:32 AM
Subject: Re: [boost] Boost.Threads, the Boost.Build system and GCC

>
> >Do you want a user-settable variable that contains a bunch of
> >compiler-specific flags, but produces no subvariant directories?
>
> Yes
>
> >It would be next to impossible to ensure link compatibility that way. If
> >your variable changed slightly between invocations of the build system,
> you
> >could find yourself trying to link together completely different targets.
>
> I think that's up to the user to ensure: if they inject options that break
> link compatiblity that's they're business, but if they just want to tweek
> optimisation options,

Tweaking optimization is easy: -sBUILD="release <optimization>space" for
example.

> or add a -pedantic flag then that's fine.

Wouldn't it be better to have a feature that they could use
with -sBUILD="debug <warning-level>pedantic" ?

Anyway, it would be trivial to add a user flags variable to the
command-lines of all toolset descriptions. I wouldn't mind doing it, but I
worry that it would effectively discourage people from writing a
corresponding feature (which is quite trivial to do, really) and the system
would degenerate into a pile of separate build scripts for different
compilers. Will people use this mechanism simply because they know their
compiler's command-line options and can't be bothered to learn to use the
build system?

-Dave