On Thu, 2013-07-04 at 02:24 +0200, Karl Wiberg wrote:
> On Thu, Jul 4, 2013 at 2:09 AM, Waskiewicz Jr, Peter P
> <[email protected]> wrote:
> 
> > On Wed, 2013-07-03 at 22:57 +0200, Karl Wiberg wrote:
> >
> > > One design that leaps to mind is to use empty patches for cover
> > > letters (with special headers in the commit message to mark it as
> > > a cover letter and to store the recipient list), and let each
> > > cover letter apply to the patches following it, until the end of
> > > the stack or the next cover letter.
> >
> > This is probably close.  Using empty patches would pollute the
> > commit indices, with stuff for StGit maintenance.  I'm not sure if
> > the upstream kernel maintainers would like this, since they'd either
> > have to manage not applying that patch, or apply patches that have
> > zero change.
> 
> Oh, we wouldn't try to make anyone else accept those empty patches:
> 
>   * stg mail would use them for cover letter generation, not mail them
>     out as patches.
> 
>   * stg commit would skip them, and only commit the "real" patches.
> 
> Several other parts of StGit would need to be taught to give the cover
> letter patches special treatment as well. I admit I haven't thought it
> through extensively, but I think it wouldn't be that hard to do.

I think this is the piece that I came to in my mind initially that would
need the work.  But just some thought on it, how about this.  Add a new
flag to stg new, something like:

$ stg new --cover <name>

That would create the new patch in the stgit refs, with whatever header
tags to flag it as a "special" patch.  That way if you wanted to come
back later to add a cover letter, pop everything, add the new patch,
then push everything.

I actually don't think this would be a heavy lift at all.

> Of course, one could consider alternative solutions that look the same
> in the StGit UI, but don't actually create the empty commits, storing
> their data elsewhere instead. However, the empty-commits solution has
> the advantage that even tools that haven't been taught to understand
> it (such as gitk) can see and manipulate the cover letters in a useful
> fashion.

Wouldn't gitk need the patches to be in the git log to see them, not
just the stgit refs?  If it does, I'm not sure we'd want to do that.

Cheers,
-PJ
_______________________________________________
stgit-users mailing list
[email protected]
https://mail.gna.org/listinfo/stgit-users

Reply via email to