Re: [PATCH 0/3] line-log: integrate -L with the standard log output pipeline
- From
Michael Montalbo <mmontalbo@gmail.com>
- Date
- May 24, 2026, 02:00 UTC
- Message-ID
- <CAC2QwmK_aFoeb1dybvkXxO+ZsQ_NoS8wUtYPvKD_=gxgcOEkCw@mail.gmail.com>
- In-Reply-To
- <CALnO6CBh7nDCwT=u1xSN2c6_x88t_gNfAaT_B4PzYKr=5i_bNA@mail.gmail.com>
On Fri, May 22, 2026 at 11:46 AM D. Ben Knoble <ben.knoble@gmail.com> wrote:
> > Hi Michael, >
Hi Ben,
Thanks for the thorough review!
Show 41 quoted lines
> On Tue, Apr 28, 2026 at 12:06 AM Michael Montalbo via GitGitGadget <gitgitgadget@gmail.com> wrote: >> >> Since its introduction, git log -L has short-circuited from >> log_tree_commit() into its own output function, bypassing log_tree_diff() >> and log_tree_diff_flush(). This skips no_free save/restore, >> always_show_header, diff_free() cleanup, and means that pickaxe (-S, -G, >> --find-object) and --diff-filter cannot suppress commits whose pairs are all >> filtered out, because show_log() runs before diffcore_std(). >> >> This series restructures the flow so that -L goes through the same >> log_tree_diff() -> log_tree_diff_flush() path as normal single-parent and >> merge diffs, then uses that to enable several non-patch diff formats. > > > Cleanup by itself to shrink the number of concepts in the code is already a good thing IMO, so getting additional features out of it is even nicer. > >> Patch 1: revision: move -L setup before output_format-to-diff derivation >> >> Preparatory reorder in setup_revisions(). The -L block sets a default >> DIFF_FORMAT_PATCH when no format is requested; move it before the derivation >> of revs->diff from output_format so the default is visible to that check. No >> behavior change on its own. > > > Straightforward, nice. > >> >> Patch 2: line-log: integrate -L output with the standard log-tree pipeline >> >> Rename line_log_print() to line_log_queue_pairs(), stripping it down to only >> queue pre-computed filepairs. log_tree_diff_flush() handles show_log(), >> diffcore_std(), and diff_flush(). This fixes pickaxe and --diff-filter >> suppression, and aligns the commit/diff separator with the rest of log >> output. Also rejects --full-diff, which is meaningless when filepairs are >> pre-computed. > > > At first I questioned the removal of the DIFF_FORMAT_NO_OUTPUT conditional in line_log_queue_pairs, but now that it only queues pairs it shouldn't be checking output formats. Good. > > I also noted that log_tree_diff() returns the result of log_tree_diff_flush() in the -L case, which is a bit different from the other patterns. I think the difference is that the other cases have some conditional logic around the log_tree_diff_flush cases (?) but I'm not sure. Perhaps that branch should also be looking at opt->loginfo ? >
Good catch. In practice I think they agree, since `log_tree_diff_flush()` returns 1 exactly when it calls `show_log()` which consumes loginfo, but matching the existing convention is cleaner. Will update.
Show 7 quoted lines
> Finally, I wonder if in describing the removal of the early return: > > > - Remove the early return in log_tree_commit() that bypassed > > no_free save/restore, always_show_header, and diff_free(). > > we might want to be more explicit that this is _because_ line-level diff is now handled in the regular pipeline? >
Agreed, will reword to: "Remove the early return in log_tree_commit() that is no longer needed now that -L output flows through log_tree_diff() and log_tree_diff_flush(); this restores no_free save/restore, always_show_header, and diff_free() cleanup."
> [I suppose we could, in theory, split the rejection of --full-diff to a separate prep commit, idk.j] >
It felt natural to me to put alongside the integration since --full-diff is not yet implementable with pre-computed filepairs. Happy to split it out if you feel strongly though.
Show 15 quoted lines
>> Patch 3: line-log: allow non-patch diff formats with -L >> >> Expand the allowlist to accept --raw, --name-only, --name-status, and >> --summary. These only read filepair metadata already set by the line-log >> machinery. Diff stat formats (--stat, --numstat, --shortstat, --dirstat) >> remain blocked because they call compute_diffstat() on full blob content and >> would show whole-file statistics rather than range-scoped ones. > > > Short and sweet. > > The stat formats are kind of like --full-diff, and I think they should probably all be rejected or all allowed: since the stats are based on the full-diff, it makes sense to enable them if we can also make -L + --full-diff semantically sensible. > > Otherwise, we'd need to find a way to make the stat formats scoped for -L. >
I am working on a follow up series that takes the second path you suggest: it adds a line-range filter in `diffcore_std()` that clips insertions and deletions to the tracked ranges before `compute_diffstat()` runs, so `--stat`, `--numstat`, etc. report range-scoped numbers. That series builds on top of these three patches, which is why stats remain blocked here.
For `--full-diff`, thinking about it more, the semantics would actually be well-defined: "filter commits by line range, but show the full diff for those commits." Right now, there might be a higher implementation barrier, though. The line-log machinery fuses commit filtering with diff generation, so there is no separate "full diff" to fall back to for display. I will soften the rejection to "not yet supported" rather than "incompatible," since it could be wired up if someone separates the two concerns.
Show 31 quoted lines
>> >> >> Michael Montalbo (3): >> revision: move -L setup before output_format-to-diff derivation >> line-log: integrate -L output with the standard log-tree pipeline >> line-log: allow non-patch diff formats with -L >> >> Documentation/line-range-options.adoc | 10 +- >> line-log.c | 30 ++---- >> line-log.h | 2 +- >> log-tree.c | 9 +- >> revision.c | 25 +++-- >> t/t4211-line-log.sh | 99 ++++++++++++++++--- >> t/t4211/sha1/expect.parallel-change-f-to-main | 1 - >> .../sha256/expect.parallel-change-f-to-main | 1 - >> 8 files changed, 120 insertions(+), 57 deletions(-) >> >> >> base-commit: 9f223ef1c026d91c7ac68cc0211bde255dda6199 >> Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-2094%2Fmmontalbo%2Fmm%2Fline-log-use-log-tree-diff-flush-v1 >> Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-2094/mmontalbo/mm/line-log-use-log-tree-diff-flush-v1 >> Pull-Request: https://github.com/gitgitgadget/git/pull/2094 >> -- >> gitgitgadget > > > A few other comments: > > - Tests should use test_grep; some do, but some don't. > - There is one occurrence of "sed | grep" that I wonder if we want to rewrite to avoid issues with exit status one side of the pipe? >
Will fix both these issues.
> Thanks for working on this! > > [Apologies for the unusual review format; this was easier for me at the moment than digging up the individual patches, and I don't think _most_ of the review would benefit from spreading out across multiple mails.] >
Thank you too! This review format worked fine for me :)
> -- > D. Ben Knoble