Re: [boost] [cmake] Pull request announcement

Subject: Re: [boost] [cmake] Pull request announcement
From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2018-09-27 15:01:28

On Thu, 27 Sep 2018 at 16:57, Rene Rivera <grafikrobot_at_[hidden]> wrote:
> On Thu, Sep 27, 2018 at 1:29 AM Mateusz Loskot via Boost <boost_at_[hidden]> wrote:
>> On Thu, 27 Sep 2018 at 05:18, Peter Dimov via Boost <boost_at_[hidden]> wrote:
>> > Stefan Seefeld wrote:
>> >
>> > > Indeed: While the above is a great starting point to support what I want,
>> > > I don't want to loose the ability for BPL to be built (and tested) as part
>> > > of the rest of Boost. I would thus prefer something that gives me the
>> > > ability to choose the build strategy e.g. using some command-line option,
>> > > including flag to be passed to the build (e.g., `b2 ... standalone=on`),
>> > > rather than having to modify the (build) code.
>> >
>> > Scanning for Jamroot to identify the project root is fundamental to Jam (the
>> > Perforce one). We could perhaps add something like
>> >
>> >     b2 --jamroot=.
>> >
>> > that would act as if there were an empty Jamroot at . (or equivalently if
>> > the Jamfile at . were named Jamroot.)
>>
>> Does it mean there could be two jamfile-s?
>> libs/gil/Jamroot for standalone build
>> libs/gil/Jamfile for in-Boost tree build
>> and the latter is simply ignored in normal b2 run, ie. without  --jamroot=.
>
>
> No. Such a theoretical feature would use the existing Jamfile at the location as the jamroot.
>
>> I think, this would be also very helpful to apply the Jamroot trick [1],
>> recently presented to me by Steven Watanabe,
>> to significantly speed up b2 startup procedures:
>
>
> But, yes, it would allow for that trick.
> In general I think the boost-root/Jamroot does way-way too much.
> If Boost where restructure to be modular only most of that work would unnecessary.

That sounds compelling enough to me :-) Thanks for the clarification

-- 
Mateusz Loskot, http://mateusz.loskot.net