On Wed, Apr 29, 2009 at 09:52:00AM -0500, Maynard Johnson wrote:
> Andi Kleen wrote:
> > On Wed, Apr 29, 2009 at 08:38:16AM -0500, Maynard Johnson wrote:
> >> Hi, Suravee,
> >> Most of Andi's recently posted 9 patches involve Intel processors. As the Intel/AMD sub-maintainer for the project, can you find time to review patch 2 and patches 5-9? I'll look at patches 1, 3, and 4. Note that patch 8 is dependent on patch 4.
> > No all patches are dependent on the previous ones.
> > You cannot treat them independently.
> So, you're saying if we don't accept, say, patch 3 (word wrap support for events), then patches 4-9 can't be accepted? I don't see a strict dependency there.
% wc -L ~/src2/oprofile/events/i386/nehalem/unit_masks
If you don't add word wrap support opcontrol --list-events with
Nehalem is totally unreadable.
> >> Andi, some of the patches you posted are related and others are completely independent, stand-alone patches. For future reference, please only group related patches. Also in future, please include a ChangeLog entry *with each patch*.
> > There's changelog in each patch where it made sense (in some cases
> If you look at most other patches that have been accepted, you'll see those patches included a patch to the ChangeLog file. That's what I'm talking about. The requirement for this is documented in the HACKING file.
Ah. That's easy to create with a script. I assumed you already
had one. I can repost with that.
But why anyone wants GNU style changelogs if you have CVS logs
is beyond me.
> > a single subject sentence was sufficient)
> > They're all dependent on each other the only exception is the error handling
> > patch (which can be safely dropped, but I found it very useful while
> > doing the large event files)
> It's to your advantage to group only related patches, since this usually implies an all-or-none kind of situation. I'll leave it at that.
Again they're all related, except for the optional error handling
ak@... -- Speaking for myself only.