Re: [boost] Request permission to apply patch to trunk.

Subject: Re: [boost] Request permission to apply patch to trunk.
From: Beman Dawes (bdawes_at_[hidden])
Date: 2013-10-22 11:36:27

On Mon, Oct 21, 2013 at 3:04 AM, Jürgen Hunold <jhunold_at_gmx.eu> wrote:

> Hi,
>
> On Monday, 21. October 2013 10:54:00 Vladimir Prus wrote:
> > On 20.10.2013 23:13, Beman Dawes wrote:
> > > I'd like to apply the attached patch to trunk. It is not supposed to
> break
> > > anything on an svn checkout, but we held off applying it until 1.55
> beta
> > > shipped because of concern it would be disruptive if it somehow messed
> up
> > > the build system right in the middle of getting the beta out.
> > >
> > > Thanks to Bjørn Roald for developing the patch.
> > >
> > > Without this patch, the only way to test modular-boost headers
> mechanism
> > > is
> > > to manually apply the patch yourself, and that pretty much limits
> testing.
> > >
> > > OK to apply?
>
> Yes, the patch looks fine.
>
> > as you probably know, the only reason this branch is not yet merged to
> trunk
> > is because there's no proper dependency between install and headers set
> up
> > yet. That is, you need "b2 headers" before "b2 install" will work.
> Whether
> > you prefer to merge/commit now is rather up to you.
>
> Please commit this one. It make thing easier.
>
> > FAOD, adding proper dependencies does not involve any theoretical
> > difficulty, just a bit of time.
>
> Right. And we can save some time by getting the things which already work
> into
> the repository. So we can concentrate on adapting the rest of the
> libraries.
>

That's a very important point. As the remaining conversion issues shrink,
the lack of this patch is limiting testing unnecessarily.

On Windows 7 and Ubunto Linux I've done a fresh SVN checkout of trunk and
verified bootstrap.bat, bootstrap.sh, and a b2 build work both before and
after the patch is applied. On Windows, I verified that after the patch is
applied a b2 build using an unpatched prior compilation of b2 still works.
These tests increase confidence that committing the patch to trunk now
isn't likely to cause breakage, so I'm going ahead with the svn commit.

Thanks,

--Beman