|
Boost-Build : |
Subject: Re: [Boost-build] Status of boost-build at sourceforge?
From: Steve M. Robbins (steve_at_[hidden])
Date: 2011-04-15 11:44:33
Hi Volodya,
Comments inline, below.
> On Sunday, March 27, 2011 13:25:35 Vladimir Prus wrote:
> > On Sunday, March 20, 2011 15:21:18 Steve M. Robbins wrote:
> > > On Sun, Mar 20, 2011 at 12:32:58PM +0300, Vladimir Prus wrote:
> > > > On Sunday, March 20, 2011 01:22:49 Steve M. Robbins wrote:
> > I think it is unfortunate coupling. We want Boost.Build to be usable
> > without Boost in general, or huge pile of headers in particular.
> > Do you think it would be possible to have boost-build package, that
> > includes both the .jam files, and the executable (now named bjam,
> > soon to be officially renamed)?
Yes, we could certainly produce a boost-build (binary) package again.
The main thing I would struggle with is: what to use as the source
package?
> > > When you say "official Boost.Build 2011.03", above, do you mean to say
> > > that Boost.Build will start having releases independently of Boost?
> >
> > Yes.
> >
> > > Will Boost.Build also continue to be bundled with Boost?
> >
> > Yes. Ideally, Boost will be bundled with specific Boost.Build release
> > that is known to work with that Boost version.
Here's my difficulty: Boost.Build sources come both independently and
bundled with Boost. So which should I choose to use as the source of
the boost-build binary package? In principle, one would expect to use
the independent release of Boost.Build, and then use *that* to build
Boost itself.
But if "Boost will be bundled with specific Boost.Build release that
is known to work with that Boost version", then I may be causing
problems for myself and should rather use the bundled Boost.Build to
build Boost. But if that's the case, then I may as well just use the
Boost sources as the source of Boost.Build (as we presently do). Then
I have one less source package to worry about.
For this reason, I'm strongly leaning towards continuing current
Debian practice of building everything from a Boost release. As
mentioned above, I could be persuaded to revive the separate
boost-build package; although, to be frank, no-one has complained yet
about bjam being bundled with the headers.
Regards,
-Steve
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