[PATCH] GCSE after reload - resubmit (SPECint +1.5% fp -0.5%).

[Paul Koning]

>>>>> "Gerald" == Gerald Pfeifer <gerald@pfeifer.com> writes:

 Gerald> On Wed, 25 Feb 2004, Roger Sayle wrote:
 >> Other maintainers may disagree, and should feel free to sponsor
 >> your request themselves.  I have no objections whatsoever if
 >> they're happy to do so.  If its any consolation, I'd had nearly 20
 >> patches approved to GCC before being granted CVS write access
 >> myself.  But perhaps policies have relaxed somewhat since then.

 Gerald> I think this makes sense.  Judging from the number of write
 Gerald> after approval committers who even hadn't added themselves to
 Gerald> the MAINTAINERS file, we apparently handed out CVS write
 Gerald> access to readily last year.

Not being in MAINTAINERS is certainly odd.

 Gerald> We shouldn't have a fixed limit, but at least a dozen patches
 Gerald> within a year might be a minimum (depending on size, impact,
 Gerald> and quality of the patches, of course).

I'm not sure that sort of number is reasonable.  For general
maintainers, perhaps -- but you want to allow for time to come up to
speed.  GCC is really hard...

For target maintainers, I don't see any way that such a notion makes
sense at all.  Depending on the target, there simply may not be any
need for major activity.  I don't believe it is right to deprecate a
target just because it isn't the subject of major code churn.

       paul