This is the mail archive of the
mailing list for the GCC project.
Re: Clean up dwarf2out.c
On Fri, Apr 30, 2010 at 12:53 PM, Manuel López-Ibáñez
> On 30 April 2010 11:54, Richard Guenther <email@example.com> wrote:
>> 2010/4/30 Manuel López-Ibáñez <firstname.lastname@example.org>:
>>> 2010/4/29 Anatoly Sokolov <email@example.com>:
>>>> ?This patch add rtx_to_double_int functions in the GCC and do clean up
>>>> ?Bootstrapped/regtested on x86_64-unknown-linux-gnu, OK for mainline?
>>>> ? ? ? ?* double-int.h (rtx_to_double_int): Declare.
>>>> ? ? ? ?* double-int.c: Include rtl.h.
>>>> ? ? ? ?(rtx_to_double_int): New function.
>>> Is double-int.h using rtx.h or the other way around? For modularity's
>>> sake, this function should go to rtx.h, so you should not need to add
>>> rtl.h to double-int.c.
>> It follows existing practice with tree_to_double_int.
> Given GCC's modularity that means nothing. Moreover, double-int.c may
> be part of the tree module, which is used by rtl.h interface anyway.
> That doesn't mean that double-int must depend on rtl.h.
> Either we start to get serious about this or we give up now.
Well, I have nothing against moving tree parts to tree.[ch] and
rtl parts to rtl.[ch].
Feel free to do that together with this patch or as a followup.