Re: [PATCH] Better insertion of hint and hbrp insns. SPU, sched, hint

[Vladimir Makarov <vmakarov at redhat dot com>]

Joseph S. Myers wrote:
> On Thu, 4 Sep 2008, Trevor_Smigiel@playstation.sony.com wrote:
>
>   
> > A question for the release managers.  What is the policy for checking in
> > to mainline after the change to stage 3?  I submitted the patch before
> > the deadline, is it ok to check it in?
> >     
>
> In general, maintainers of parts of the compiler have discretion to decide 
> what changes to allow in even if they do not strictly meet the general 
> definition of appropriate changes for stage 3.  Maintainers of the 
> language and target independent parts of the compiler, the C and C++ front 
> ends, the C++ runtime library, libgcc and primary and secondary targets 
> need to be especially conservative about this so that we achieve the 
> desired stabilisation and avoid new features introducing regressions; 
> similarly, we can be less conservative early in stage 3 than later on.
>
> There is no general rule that submission before stage 3 means a patch can 
> be committed during stage 3; the relevant maintainers need to consider the 
> risks and benefits to decide whether it's OK at a particular point in 
> stage 3.
>
>   
The change in machine independent part is very safe.  It is actually one 
new function which is used by one target.  So as insn scheduler 
maintainer I don't worry that the change will create a problem on 
stage3.  So it is ok to me to commit the patch even on stage 3.

Vlad