Re: [boost] Re: Anyone planning a compression library?

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-02-14 15:46:32

At 08:40 PM 2/14/2001 +0200, Peter Dimov wrote:

 >From: <jeff_at_[hidden]>
 >
 >
 >> > > use a boost library.  Unless I am missing something, boost library
 >> > > standards don't allow this kind of external library dependency
 >> > > (standard library excepted).
 >> >
 >> > Don't they?
 >>
 >> Direct from the guidelines page
 >> (http://www.boost.org/more/lib_guide.htm)
 >>
 >> -->" Do not use libraries other than the C++ Standard Library or
 >> Boost."
 >
 >I see; an interesting point. So to provide a zlib C++ frontend, a boost
 >author would need to, basically, rewrite zlib from scratch (incorporating 

 >it
 >as-is creates maintenance problems since the versions may desync.) Sounds
 >like reinventing the wheel to me - zlib is heavily tested and "just 
works."

The intent of the guideline is to ensure that Boost doesn't have the kinds 
of dependencies and couplings which plague some other other libraries.  You 
know, where you download A, which tells you that you must download B, and 
so on until the recursion finally terminates when you reach a required 
library which is no longer available for your platform and the whole house 
of cards collapses.

One possible way to deal with this (assuming zlib is of high quality, its 
authors are interested, and meets other criteria) might be to bring it 
under the Boost umbrella.  This might be done in two ways - 1) simply 
include it in the distribution as an implementation detail without directly 
exposing it, or 2) convince the authors to submit it to boost like any 
other library.

We haven't had a major library submitted that wants to continue to maintain 
itself separately, with its own CVS, web site, and so forth, but that is 
bound to happen.  We might want to be mentally prepared to deal with such a 
request.  I really don't know enough about zlib to know if it is a good 
candidate, but it may well be.

--Beman