Re: [patch] for PR 27735

[Eric Botcazou <ebotcazou at libertysurf dot fr>]

> yes (the arguments of fix_loop_placement changed, so this spot needs to
> be changed; and I noticed that in fact the fix_loop_placement calls are
> in fact not needed here, since remove_path calls fix_loop_placement if
> necessary).

Err... no, only the return value of fix_loop_placement has been changed.
If that was the main argument for removing the calls, please put them back, 
provided the fix still works with them.

> Is there some document specifying how the testcases should look like?
> Otherwise, it does not make much sense to me to require one ad-hoc
> convention over another ad-hoc convention.

Joseph answered, but http://gcc.gnu.org/codingconventions.html explicitly 
mentions the header for testcases.

> I will send updated patch once it is tested.

Thanks in advance.

-- 
Eric Botcazou