From: Ross Smith (r-smith_at_[hidden])
Date: 2002-02-25 11:33:03
Peter Dimov wrote:
> From: "Ross Smith" <r-smith_at_[hidden]>
> > Because, after examining the matter carefully, I've come to the
> > conclusion that duplicating the functionality of the parts of Boost I
> > find useful in my own code is a more efficient use of my time than
> > trying to make Boost usable.
> Which parts of Boost do you find useful, from a functionality standpoint?
Let's see ... The biggest one is regex. At the moment I'm using a C++
wrapper around the Posix <regex.h> facilities (my code doesn't need to
be portable to non-Posix systems), which does what I want but (being C)
is very unsatisfactory from the internationalisation point of view. I'm
currently thinking that my best approach would be to extract the regex
library from boost, remove the boost dependencies, and distribute it
myself, since it already has a pretty good confifgure/install system of
(I'll probably end up doing the same thing with Spirit, which isn't
officially part of boost yet but is intimately dependent on it.)
I haven't used the graph library yet, but I can see needing it in future
The only other parts of boost I'd have liked to be able to use are
<boost/integer.hpp> (trivial to duplicate the functionality) and the
function object stuff (bind, compose, etc.; not quite so trivial but
still not too difficult). Most of the other useful stuff I've replaced
anyway, since I had problems with the way the boost version worked,
aside from the installation issue (static_assert, type_traits, thread,
> Why do you consider them so unusable that duplicating is more efficient use
> of your time than making them work?
Because there's no way for end users to install them.
> How can they be improved to avoid this?
By making it easy to install the whole thing.
> Please don't speak about Boost in general. This gives us little to no
But boost in general is the problem, not any one specific library. The
fundamental problem is the lack of an easy way for non-developers to
install the whole thing.
Approaching this on a piece-by-piece basis is a waste of time because
(1) the individual sub-libraries have too many dependencies on each
other, and (2) developing the same facilities over and over again for
each part, instead of just once for the whole thing, would be silly.
-- Ross Smith ...................................... Auckland, New Zealand r-smith_at_[hidden] ......................... http://storm.net.nz/~ross/ "We need a new cosmology. New gods. New sacraments. Another drink." -- Patti Smith
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk