Re: [Boost-build] Future development

From: Michael Caisse (boost_at_[hidden])
Date: 2007-10-18 16:36:33

David Abrahams wrote:
> on Wed Oct 17 2007, Vladimir Prus <ghost-AT-cs.msu.su> wrote:
>
>   
>> I start to wonder if this thread has anything to do with Boost.Build any longer?
>> It seems purely a discussion of how to do something with CMake,
>> and therefore not a topic of this list.
>>
>> Am I wrong?
>>     
>
> I think so.  We're discussing how to implement the Boost.Build
> abstractions on top of an engine different from bjam.
>
>   
I thought this might come up eventually. 

So we are talking about the boost community supporting the Boost.Build
abstraction. This abstraction sits on top of an engine. The only argument
I have really heard thus far concerning a move to a new engine is that
somebody else will support it.

So the question is (and it is an honest question) are there so many bugs
and problems with the current engine (bjam) that it warrants a re-write
of Boost.Build for a new engine?

The second complaint I have heard revolves around the current Boost.Build
abstraction being hard to understand and make updates to. That may be the case;
however, throwing out a "working" system for a newly designed system is a
typical blunder (that I have been guilty of myself)
< http://www.joelonsoftware.com/articles/fog0000000069.html >. I suspect that
there is a lot of knowledge wrapped up into the current Boost.Build abstraction.

If the engine is stable and the current abstraction implementation just 
hard to support, what makes the new abstraction for a different engine
not have this problem? Better documentation? Better coding style?
Community reviews of the work?

I'm not bashing change, but I'm wondering if there are some "rose tinted
glasses". Lets take a real example: Gennadiy has questions about supporting
usage requirements on a per header basis. This is something supported by
the Boost.Build abstraction. Currently Voloyda is responding to the email
thread. If Gennadiy had the same questions but the engine was CMake, I
assume Doug (or delegate) would be answering the questions. Where is there
a savings in Boost community effort?

Best Regards -
Michael

-- 
----------------------------------
Michael Caisse
Object Modeling Designs
www.objectmodelingdesigns.com