Should 3.2 still search backward/ headers by default?

[Gabriel Dos Reis]

Paolo Carlini <pcarlini@unitus.it> writes:

| Phil Edwards wrote:
| 
| >Not strong ones.  Just weaker ones (simplification, removal of unneeded code,
| >etc).  I'm happy to see them searched by default, but happier to see them
| >only present when the user explicitly requests pre-Standard stuff.
| >
| Personally, I'm tempted to be with Phil, but weeks ago, when some 
| backwards bits (backward/hash_map, I believe) turned out to be broken 
| due to the ext/ work, *two* people (Jakub Jelinek and Philipp Thomas) 
| noticed it at the same time! Therefore, I'm afraid a few widespread 
| packages still use them heavily :-(


My opinion is to keep the backward/ bits in not because I'm a fun of
them, because I don't think we can just get rid of them after two
releases of the 3.x series.  People still use GCC-2.95 because they
feel GCC-3.x is not yet stable nor usable.  Once we gain at least the
claimed stability of GCC-2.95 backward/ removal won't wake emotions.

-- Gaby