Re: [Boost-build] Future development

From: Rene Rivera (grafikrobot_at_[hidden])
Date: 2007-10-03 01:19:31

Vladimir Prus wrote:
> Rene Rivera wrote:
> 
>> I looked, and... I'm not sure how much of it I can use. It relies fairly
>> heavily on Boost libraries which presents a rather annoying bootstrap
>> problem. 
> 
> Oooh! I though it only uses external hash library, but otherwise self-contained.
> It's indeed a problem.

AFAICT it uses: enable_if, type_traits, ref, assert, aligned_storage, 
intrusive_ptr (and related files), noncopyable, array, config, format, 
shared_ptr, iterators, value_init, and ptr_container.

>> It also goes against my plan to star using C++ incrementally 
>> since it's a ground-up rewrite.
> 
> Well, assuming it works it's OK; unfortunately we don't have much
> unit tests for bjam, so it's hard to be sure there are no bugs,
> and fixing the bugs with a completely changed codebase might
> be much harder.

Yep. Especially since some the recent code change to bjam are much 
better structured since I abandoned any hope of keeping in sync with 
Perforce/Jam. Given that Perforce isn't maintaining Jam any more AFAICT. 
  And along the way I've been adding test cases as bugs get fixed. 
Probably for me the biggest core change I don't see how do easily is 
redoing the parser.

So if any one has ideas on how to solve the problem of generating C++ 
based parser with minimal fuzz please speak up :-) (And by minimal I 
mean not requiring a large chunk of external code or tools)

-- 
-- Grafik - Don't Assume Anything
-- Redshift Software, Inc. - http://redshift-software.com
-- rrivera/acm.org - grafik/redshift-software.com
-- 102708583/icq - grafikrobot/aim - grafikrobot/yahoo