Re: [PATCH] fix -Werrors from new readline

[Daniel Jacobowitz <drow@mvista.com>]

On Sun, Dec 08, 2002 at 11:34:02PM -0500, Elena Zannoni wrote:
> Daniel Jacobowitz writes:
>  > On Sun, Dec 08, 2002 at 08:23:40PM -0500, Elena Zannoni wrote:
>  > > 
>  > > 
>  > > And the makefile.
>  > > 
>  > > 2002-12-08  Elena Zannoni  <ezannoni@redhat.com>
>  > > 
>  > > 	* Makefile.in (readline_h): Define.
>  > > 	(completer.o): Depend on readline_h.
>  > > 	(corelow.o): Ditto.
>  > > 	(event-top.o): Ditto.
>  > > 	(exec.o): Ditto.
>  > > 	(solib.o): Ditto.
>  > > 	(source.o): Ditto. 
>  > > 	(symfile.o): Ditto. 
>  > > 	(symmisc.o): Ditto. 
>  > > 	(top.o): Ditto.
>  > > 	(tracepoint.o): Ditto. 
>  > > 	(utils.o): Ditto.
>  > > 	(cli-dump.o): Ditto.
>  > > 	(tui-hooks.o): Ditto. 
>  > > 	(tuiWin.o): Ditto. 
>  > 
>  > > +readline_h = 	$(READLINE_SRC)/readline.h
>  > 
>  > Erk, is this necessary?  It comes from the combined source tree, but
>  > it's basically a 'system' header.  I compile using the system readline
>  > instead of the included one; this is just another thing which would get
>  > in the way of doing that.
>  > 
> 
> Hmm, but what happens if the system readline and gdb's readline get
> out of sync again?  BTW, there was a readline_h already used in some
> tui files, but it wasn't defined.

I cry? :)  I've been maintaining patches to use the system readline. 
You've just integrated pretty much all of them.

> Rereading your sentence... you are trying to get rid of src/readline?

Certainly not.  What I want to do is add a --with-system-readline,
analagous to GCC's --with-system-zlib; it's pretty straightforward to
do as long as the interface stays "reasonably" stable.

Upon rethinking it, I doubt the $(readline_h) dependency will be a
problem.  Someday I would like to make GDB use BFD's make dep-am
mechanism...

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer