Re: [RFC PATCH 10/10] pack-objects: improve partial packfile reuse

From: Jeff King <peff@peff.net>
Date: Oct 2, 2019, 15:57 UTC
Message-ID: <20191002155721.GD6116@sigill.intra.peff.net>
In-Reply-To: <xmqqy2yrbmqu.fsf@gitster-ct.c.googlers.com>

On Fri, Sep 13, 2019 at 08:06:01PM -0700, Junio C Hamano wrote:

Show 27 quoted lines

> Jeff King <peff@peff.net> writes:
> 
> > On Fri, Sep 13, 2019 at 03:29:00PM -0700, Junio C Hamano wrote:
> >
> >> This comment has nothing to do with the change, but the way the
> >> patch is presented is quite hard to follow, in that the preimage or
> >> the common context lines do not help understand what the new code is
> >> doing at all ;-)
> >> 
> >> I'll come back to the remainder of the patch later.  Thanks.
> >
> > I applaud Christian's effort to tease it out into separate patches.
> 
> Ah, no question about it.  I have a suspicion that 10/10 alone may
> still be a bit too large a ball of wax, but with all the earlier
> preparatory steps are bite-sized and trivial to see how they are
> correct.
> 
> The "way the patch is presented" comment was not at all about what
> Christian did, but was about what the diff machinery computed when
> comparing the 9th step Christian created and the final step.  In its
> attempt to find and align common context lines, it ended up finding
> blank lines and almost nothing else in the earlier part of the
> patch, not just making it harder to read the new helper function
> (i.e. the best way to read record_reused_object(), for example, is
> to look only at '+' and ' ' lines, because the '-' lines are
> irrelevant), it also made it hard to see what got discarded.

Hmm, I see the early parts of this graduated to 'next'. I'm not sure everything there is completely correct, though. E.g. I'm not sure of the reasoning in df75281e78 (ewah/bitmap: always allocate 2 more words, 2019-09-13).

I think the "block+1" there is actually because "block" might be "0". Prior to 2820ed171a (ewah/bitmap: introduce bitmap_word_alloc(), 2019-09-13) from the same series, that could never be the case because we know that we always start with at least 32 allocated words. But after that we _could_ start with an empty word array, and allocating "block * 2" would not make forward progress.

And then the "2 more words" thing is used as justification in the next patch, 04a32d357f (pack-bitmap: don't rely on bitmap_git->reuse_objects, 2019-09-13). I won't say that there isn't some subtle dependency there, but I certainly don't remember any logic like that at all. ;) So I think it might bear a little more scrutiny.

I'm sorry for being so slow on giving it a more careful review. I was traveling for work, then playing catch-up, and am now going on vacation. So it might be a little while yet.

-Peff