Re: trunk patch: moving gimple_seq_node from coretypes.h to gimple.h

[Basile Starynkevitch <basile at starynkevitch dot net>]

On Tue, 2010-06-22 at 22:04 +0200, Steven Bosscher wrote:
> On Tue, Jun 22, 2010 at 8:30 PM, Basile Starynkevitch
> <basile@starynkevitch.net> wrote:
> > Hello All,
> >
> > since gimple_seq_node is not very useful in coretypes.h I bootstrapped
> > the following attached patch to trunk rev.161214 on
> > x86_64-unknown-linux-gnu with c,c++,lto languages.
> >
> > gcc/ChangeLog entry:
> >
> > 2010-06-22  Basile Starynkevitch  <basile@starynkevitch.net>
> >        * coretypes.h (gimple_seq_node_d, gimple_seq_node)
> >        (const_gimple_seq_node): Removed typedefs.
> >
> >        * gimple.h  (gimple_seq_node_d, gimple_seq_node)
> >        (const_gimple_seq_node): Added typedefs moved from coretypes.h.
> 
> Please do not do this. gimple_seq is used in target.h for example, but
> target.h should not have to include gimple.h.
> 
> This patch is a step in the wrong direction.

But the patch did not remove the gimple_seq typedef, only the
gimple_seq_node one. So I don't understand why you believe it is a wrong
step.

Why would target.h need gimple_seq_node without including gimple.h? BTW,
the proposed patch did bootstrap, so I suppose that some code did
compile correctly and included target.h.

[Most passes work on gimple_seq, not gimple_seq_node].

Cheers.

-- 
Basile STARYNKEVITCH         http://starynkevitch.net/Basile/
email: basile<at>starynkevitch<dot>net mobile: +33 6 8501 2359
8, rue de la Faiencerie, 92340 Bourg La Reine, France
*** opinions {are only mines, sont seulement les miennes} ***