Re: [PATCH 2/2] bloom: enable multiple pathspec bloom keys

From: Junio C Hamano <gitster@pobox.com>
Date: Jul 1, 2025, 15:19 UTC
Message-ID: <xmqqo6u4kkg0.fsf@gitster.g>
In-Reply-To: <C8E0D62E-11B1-4921-AD4C-2905F10E07B6@gmail.com>

Lidong Yan <yldhome2d2@gmail.com> writes:

Show 11 quoted lines

> Junio C Hamano <gitster@pobox.com> writes:
>> Totally outside the topic, but I wonder if we can further optimize
>> by adding an early rejection using .nowildcard_len?  Instead of
>> allowing a wildcarded "dir/*" pathspec element from disabling the
>> Bloom filter altogether, we could say "dir/ is not possibly altered,
>> so there may be dir/A, dir/B, etc., in the directory, nothing that
>> would match dir/* wildcard would have been modified", couldn't we?
>
> I think, except for PATHSPEC_EXCLUDE, all other pathspec magic flags
> could potentially be optimized using .nowildcard_len by restricting checks to
> just the dir/ part of each pathspec item.

A good observation.

I do not know about icase; though. Asking about "Dir/Path" and getting "Dir/ or Dir/Path cannot possibly be in the set of paths that were modified" from the changed-path Bloom filter would not help us optimize the tree comparison out, when we do not want to miss modifications for "dir/path".

Show 8 quoted lines

> Here;s are all possible pathspec magic
> #define PATHSPEC_FROMTOP	(1<<0)
> #define PATHSPEC_MAXDEPTH	(1<<1)
> #define PATHSPEC_LITERAL	(1<<2)
> #define PATHSPEC_GLOB		(1<<3)
> #define PATHSPEC_ICASE		(1<<4)
> #define PATHSPEC_EXCLUDE	(1<<5)
> #define PATHSPEC_ATTR		(1<<6)