Re: Patch 6/9: remove "allocno" from live_range_t

[Jeff Law <law at redhat dot com>]

On 06/24/10 17:37, Bernd Schmidt wrote:
> On 06/18/2010 06:40 PM, Jeff Law wrote:
>    
> > On 06/18/10 08:09, Bernd Schmidt wrote:
> >      
> > > This is another patch to remove "allocno" from names, this time involving
> > > live ranges.  Subsequent patches will change IRA to associate live ranges
> > > with other objects than allocnos; even without that motivation I believe
> > > that the code becomes more readable due to the shorter identifiers.
> > > There's
> > > really no need to call a live range an allocno_live_range.
> > >
> > >      * ira-int.h (struct live_range, live_range_t): Renamed from struct
> > >      ira_allocno_live_range and allocno_live_range_t; all uses changed.
> > >      * ira-build.c (live_range_pool): Renamed from
> > > allocno_live_range_pool.
> > >      All uses changed.
> > >        
> > OK
> >      
> Thanks for the OKs.  I've committed patches 1-6; I think I'll wait for a
> conclusion about whether the final piece should go in before I commit
> the conversion to objects (unless there's consensus that it's a good
> idea on its own).
>    
NP.

I haven't looked at the meat of the final piece (and can't until next 
week), but I know that handling double-word allocations better is a 
serious issue in IRA, perhaps the most serious issue from a code 
generation standpoint.   The direction taken so far seems quite 
reasonable to me.

jeff