Re: [commit] Move hex_string et.al. to defs.h / utils.c

[Andrew Cagney <cagney@gnu.org>]

>>> Somehow some some #include "language.h" were lost with PaulH's recent 
>>> *_string_custom cleanup.
> 
> 
>>> Rather than add lots of those includes, I moved the hex_string and 
>>> hex_string_custom functions to utils.c, and the corresponding 
>>> declaration to defs.h.  I also took the oportunity to improve the code a 
>>> little switching it to get_cell().
> 
> 
> Andrew,
> 
> Actually, the #includes were dropped deliberately, and you've done the piece
> what was SUPPOSED to happen and somehow got dropped: moving the decls to 
> defs.h and defns to utils.c.  Thanks.

"oops" :-)

> I must be careful not to overlook
> mere warnings (speaking of which, it would be a nice goal to remove all
> warnings from GDB compilations, lest they distract).  

You should always configure with --enable-gdb-warnings=,-Werror 
(perhaphs GDB should automatically configure in -Werror when there's a 
CVS repository say).  See:
http://sources.redhat.com/gdb/current/onlinedocs/gdbint_13.html#SEC119
You'll also want to check out gdb_mbuild.sh.

>>> To answer your question about my thoughts on get_cell - I think it sux 
>>> less than the old code's local static buffer :-)
> 
> 
> It does make one long for fast GC, does it not?  Or perhaps something 
> region-based, where the region gets deallocated at, say, a point in the
> top-level command loop.

Garbage collection, yes.

Andrew