On Tue, Jan 08, 2019 at 09:39:48AM -0800, Junio C Hamano wrote:
> > I skimmed them; they look good to me. 6 and 8 are particularly
> > satisfying; getting rid of hash copy operations just feels nice. :)
> >
> > Junio only took 1 to 5 into pu; 6, 7 and its sidekick 8, 10 and 11
> > conflict with sb/more-repo-in-api; 9 could go in unmodified.
>
> I think these later steps are based on something a lot newer than
> the result of applying your updates to the jk/loose-object-cache
> series they fix. I think I untangled and backported one of the
> earlier commits but then I stopped after 05/11.
Sorry, I applied René's patches on top of master and then built on that,
treating it like a new topic. But of course your jk/loose-object-cache
is older.
> I do not think it is important to keep jk/loose-object-cache and
> these two follow-up topics mergeable to the maintenance track, so
> I'll see if the patches behave better if queued directly on top of
> 3b2f8a02 ("Merge branch 'jk/loose-object-cache'", 2019-01-04), or
> even a yet newer random point on 'master'.
Yeah, they should. I think one of them will need René's patch, which
changes the body of quick_has_loose(). I can roll it as a separate topic
if that's easier (or just wait a week or so until René's cleanups
graduate).
-Peff
