Re: [PATCH v4 5/6] last-modified: check pathspec against Bloom filter first

From: Patrick Steinhardt <ps@pks.im>
Date: Sep 10, 2026, 07:04 UTC
Message-ID: <aqJWb1kq81A8AJWI@pks.im>
In-Reply-To: <20260901-toon-speed-up-last-modified-v4-5-a09949800404@iotcl.com>

On Tue, Sep 01, 2026 at 11:10:25AM +0200, Toon Claes wrote:

Show 24 quoted lines

> When git-last-modified(1) starts, it builds a list of all the paths
> matching the pathspec it needs to find the last modifying commit for.
> For example, every file and subdirectory listed by:
> 
>     $ git last-modified -t --max-depth=0 -- src/
> 
> As it resolves a commit for each path during the revision walk, it drops
> that path from the list.
> 
> To avoid diffing trees for every commit, Bloom filters are used when
> available. For each remaining path, the commit's Bloom filter is checked
> to see whether the commit changed that path. The Bloom filter says
> either "no" or "maybe", and only in the latter case is the diff
> calculated.
> 
> git-log(1) does this differently. It does not expand the pathspec but
> checks the Bloom filter against the pathspec itself. This way, commits
> not touching any path matching the pathspec can be discarded as a whole.
> 
> Apply this same check to git-last-modified(1). In a previous commit the
> function revs_maybe_changed_in_bloom(), used by git-log(1), was made
> public. Use this as a pre-filter in git-last-modified(1). After this
> pre-filter, paths are still checked one-by-one to only find those which
> don't have a "last commit" yet.

So in theory, we _might_ now do some of the checks multiple times. But the expectation is that the number of pathspecs is typically much lower than the number of expanded paths to check against, so in most cases it should be faster to do this pre-filtering?

It'll probably be possible to craft edge cases where the new logic is slower because we now do more work in the matching case. But overall I think this is a sensible tradeoff. After all, we use the same tradeoff in git-log(1).

Show 12 quoted lines

> With `--show-trees` the list holds more than the paths matching the
> pathspec. It also holds each parent tree entry, up to the root. Each of
> those can resolve to a different commit. Thus for the pathspec "a/b/c",
> the list will also hold "a" and "a/b".
> 
> When a commit touches "a/other", that commit could be the last commit
> for "a", but revs_maybe_changed_in_bloom() would discard it, because it
> doesn't match the full pathspec.
> 
> Instead, when `--show-trees` is given, use
> revs_maybe_changed_in_bloom_with_parents(), which indicates the commit
> maybe changed any of the paths leading up to the path in the pathspec.

You explain what we do and why it's safe, which is good. But what's missing is the "why". As far as I understand the reason is performance, but if so I'd have expected a benchmark demonstrating the benefit.

Patrick