Re: [boost] Modularization criteria (was Re: Boost Modularization: did we get it right?)

Subject: Re: [boost] Modularization criteria (was Re: Boost Modularization: did we get it right?)
From: Dave Abrahams (dave_at_[hidden])
Date: 2012-05-21 12:36:17

on Sat May 12 2012, "Vicente J. Botet Escriba" <vicente.botet-AT-wanadoo.fr> wrote:

> Le 08/05/12 13:44, Dave Abrahams a écrit :
>> Hi All,
>>
>> As we head toward a modularized Boost, Daniel Pfeifer we on the Ryppl
>> project would like confirmation that we've correctly (or at least
>> sensibly, when there's no obvious "correct") identified the module
>> boundaries in Boost's monolithic SVN repository.  If library authors
>> could take a few moments to examine the contents of your library's repo
>> at https://github.com/boost-lib, and let us know, we'd be most grateful.
>>
>> Thanks!
>>
> Hi,
>
> I guess that the answers to these question could be found somewhere,
> but as we are discussing about whether the split is right or not I
> would like to have some questions answered explicitly.
>
> Which have been the criteria to split the Boost libraries in modulus?

Mostly the split has been along the lines of categorization as separate
libraries in the library documentation (i.e. "utility" is all one
library).  However, we've begun making some lower-level split-ups.

> Could the dependencies between modulus contain cycles or should we
> avoid them?

You should avoid them.  If avoiding them is impossible, we have a way to
deal with it, but in general avoiding them means splitting up some
libraries into separate parts, like the part that can be header-only and
the part that needs binaries.

> Do we have a tool to get the dependencies direct or indirect from a
> modulo?

https://github.com/ryppl/boost-modularize/blob/master/metadatize.py is
what we're using to generate correct dependency information in
CMakeLists.txt files.

-- 
Dave Abrahams
BoostPro Computing
http://www.boostpro.com