From: Jeff King <peff@peff.net>
To: Taylor Blau <me@ttaylorr.com>
Cc: git@vger.kernel.org, jacob@initialcommit.io, gitster@pobox.com
Subject: Re: [PATCH 5/7] shortlog: implement `--group=author` in terms of `--group=<format>`
Date: Mon, 10 Oct 2022 22:15:32 -0400 [thread overview]
Message-ID: <Y0TRxDRFRgFonhtH@coredump.intra.peff.net> (raw)
In-Reply-To: <Y0TLf/J22ioQ5UCt@nand.local>
On Mon, Oct 10, 2022 at 09:48:47PM -0400, Taylor Blau wrote:
> I think that shortlog_finish_setup() is probably the most reasonable
> choice (and I changed it to that locally). I spent a day or so thinking
> on and off about this while writing the series, and didn't come up with
> any satisfying names.
>
> I think it points to something deeper about the API that doesn't quite
> sit right with me. But shortlog_finish_setup() is the least-unsatisfying
> name so far, so let's go with that ;-).
Yeah, touching that block in make_cover_letter() definitely tickled my
memory that there was some awkwardness there. That's when I added
shortlog_init() at all (which is good, because otherwise you'd have had
an uninitialized log.format string-list).
I think the general pattern of:
foo_init();
foo.option = whatever;
foo_finish_setup();
foo_do_the_thing();
is OK. It's a little complicated, but it gives callers the freedom to
tweak options with a lot of flexibility (e.g., based on command line
config, command line options, etc). We have a similar pattern with
diff_setup_done().
The other option is for any option-tweaking to go through methods which
maintain invariants (like if GROUP_AUTHOR is set, then the "%an" format
has been added). That's usually the "right" object-oriented way to do
it. But I think even in this simple case it gets awkward, because the
correct format can't be known until you see whether log->email is set.
So it really has to be a "finalize" step after both log->email and
log->group are set.
> > This loses the HAS_MULTI_BITS() optimization. The idea there is that if
> > you have a single group-by that can't produce multiple outputs, then
> > there's no need to do duplicate detection.
> >
> > The equivalent in an all-formats world is something like:
> >
> > log.format.nr > 1 && !log.trailers.nr
> >
> > (because trailers are special in that one trailer key can produce
> > multiple idents for a single commit).
>
> Hmm. I suspect that this is going to become more complicated by the time
> that you read the final patch ;-). Let's wait until then and see what
> you think.
Yes, it's pretty gross with the trailer util thing. You'd probably want
to do something like:
static int needs_dedup(const struct string_list *formats)
{
struct string_list_item *item;
if (formats->nr > 1)
return 1;
for_each_string_list_item(item, formats) {
if (item->util)
return 1;
}
return 0;
}
and perhaps call it only once and cache the result, rather than
iterating for each commit/format.
But if we leave trailers as is, then the logic is much easier. And
implementing the optimization for --group=format should fall out pretty
naturally (and that both preserves it for --group=author, and extends it
to --group=format, which I should have noticed when reviewing patch 4).
> > > @@ -439,8 +440,8 @@ int cmd_shortlog(int argc, const char **argv, const char *prefix)
> > > log.file = rev.diffopt.file;
> > > log.date_mode = rev.date_mode;
> > >
> > > - if (!log.groups)
> > > - log.groups = SHORTLOG_GROUP_AUTHOR;
> > > + shortlog_init_group(&log);
> > > +
> > > string_list_sort(&log.trailers);
> >
> > Now that we have a "finish the shortlog init" function, probably this
> > trailer sort should go into it, too. The current caller doesn't need it,
> > but it removes on more gotcha from using the shortlog API.
>
> We'll drop this list by the end of the series, too, so it probably isn't
> worth moving it into shortlog_finish_setup() in the interim.
Ah, right. Well, see my other comments. :)
-Peff
next prev parent reply other threads:[~2022-10-11 2:15 UTC|newest]
Thread overview: 72+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-10-11 0:33 [PATCH 0/7] shortlog: introduce `--group=<format>` Taylor Blau
2022-10-11 0:34 ` [PATCH 1/7] Documentation: extract date-options.txt Taylor Blau
2022-10-11 0:54 ` Jeff King
2022-10-11 1:02 ` Taylor Blau
2022-10-11 0:34 ` [PATCH 2/7] shortlog: accept `--date`-related options Taylor Blau
2022-10-11 0:48 ` Jeff King
2022-10-11 1:14 ` Taylor Blau
2022-10-11 0:34 ` [PATCH 3/7] shortlog: extract `--group` fragment for translation Taylor Blau
2022-10-11 10:55 ` Ævar Arnfjörð Bjarmason
2022-10-11 13:24 ` Jeff King
2022-10-11 0:34 ` [PATCH 4/7] shortlog: support arbitrary commit format `--group`s Taylor Blau
2022-10-11 1:12 ` Jeff King
2022-10-11 1:24 ` Taylor Blau
2022-10-11 1:28 ` Taylor Blau
2022-10-11 2:03 ` Jeff King
2022-10-11 2:02 ` Jeff King
2022-10-21 2:39 ` Taylor Blau
2022-10-21 5:21 ` Jeff King
2022-10-21 22:12 ` Taylor Blau
2022-10-11 0:34 ` [PATCH 5/7] shortlog: implement `--group=author` in terms of `--group=<format>` Taylor Blau
2022-10-11 1:34 ` Jeff King
2022-10-11 1:48 ` Taylor Blau
2022-10-11 2:15 ` Jeff King [this message]
2022-10-21 2:03 ` Taylor Blau
2022-10-21 2:02 ` Taylor Blau
2022-10-21 5:03 ` Jeff King
2022-10-21 22:05 ` Taylor Blau
2022-10-11 10:57 ` Ævar Arnfjörð Bjarmason
2022-10-11 11:07 ` Ævar Arnfjörð Bjarmason
2022-10-11 0:34 ` [PATCH 6/7] shortlog: implement `--group=committer` " Taylor Blau
2022-10-11 1:35 ` Jeff King
2022-10-11 0:34 ` [PATCH 7/7] shortlog: implement `--group=trailer` " Taylor Blau
2022-10-11 1:50 ` Jeff King
2022-10-11 1:57 ` Jeff King
2022-10-21 1:38 ` Taylor Blau
2022-10-21 3:11 ` [PATCH v2 0/6] shortlog: introduce `--group=<format>` Taylor Blau
2022-10-21 3:11 ` [PATCH v2 1/6] shortlog: accept `--date`-related options Taylor Blau
2022-10-21 5:32 ` Jeff King
2022-10-21 21:55 ` Taylor Blau
2022-10-21 3:11 ` [PATCH v2 2/6] shortlog: make trailer insertion a noop when appropriate Taylor Blau
2022-10-21 5:38 ` Jeff King
2022-10-21 21:57 ` Taylor Blau
2022-10-21 3:11 ` [PATCH v2 3/6] shortlog: extract `--group` fragment for translation Taylor Blau
2022-10-21 5:40 ` Jeff King
2022-10-21 3:11 ` [PATCH v2 4/6] shortlog: support arbitrary commit format `--group`s Taylor Blau
2022-10-21 5:48 ` Jeff King
2022-10-21 22:07 ` Taylor Blau
2022-10-21 3:11 ` [PATCH v2 5/6] shortlog: implement `--group=author` in terms of `--group=<format>` Taylor Blau
2022-10-21 5:50 ` Jeff King
2022-10-21 3:11 ` [PATCH v2 6/6] shortlog: implement `--group=committer` " Taylor Blau
2022-10-21 5:51 ` Jeff King
2022-10-21 5:53 ` [PATCH v2 0/6] shortlog: introduce `--group=<format>` Jeff King
2022-10-21 22:25 ` [PATCH v3 0/7] " Taylor Blau
2022-10-21 22:25 ` [PATCH v3 1/7] shortlog: accept `--date`-related options Taylor Blau
2022-10-21 22:25 ` [PATCH v3 2/7] shortlog: make trailer insertion a noop when appropriate Taylor Blau
2022-10-21 22:25 ` [PATCH v3 3/7] shortlog: extract `--group` fragment for translation Taylor Blau
2022-10-21 22:25 ` [PATCH v3 4/7] shortlog: support arbitrary commit format `--group`s Taylor Blau
2022-10-22 0:28 ` Jeff King
2022-10-21 22:25 ` [PATCH v3 5/7] shortlog: extract `shortlog_finish_setup()` Taylor Blau
2022-10-21 22:25 ` [PATCH v3 6/7] shortlog: implement `--group=author` in terms of `--group=<format>` Taylor Blau
2022-10-21 22:25 ` [PATCH v3 7/7] shortlog: implement `--group=committer` " Taylor Blau
2022-10-22 0:31 ` [PATCH v3 0/7] shortlog: introduce `--group=<format>` Jeff King
2022-10-24 18:55 ` [PATCH " Taylor Blau
2022-10-24 18:55 ` [PATCH 1/7] shortlog: accept `--date`-related options Taylor Blau
2022-10-24 18:55 ` [PATCH 2/7] shortlog: make trailer insertion a noop when appropriate Taylor Blau
2022-10-24 18:55 ` [PATCH 3/7] shortlog: extract `--group` fragment for translation Taylor Blau
2022-10-24 18:55 ` [PATCH 4/7] shortlog: support arbitrary commit format `--group`s Taylor Blau
2022-10-24 18:55 ` [PATCH 5/7] shortlog: extract `shortlog_finish_setup()` Taylor Blau
2022-10-24 18:55 ` [PATCH 6/7] shortlog: implement `--group=author` in terms of `--group=<format>` Taylor Blau
2022-10-24 18:55 ` [PATCH 7/7] shortlog: implement `--group=committer` " Taylor Blau
2022-10-24 21:59 ` [PATCH 0/7] shortlog: introduce `--group=<format>` Junio C Hamano
2022-10-24 23:45 ` Jeff King
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
List information: http://vger.kernel.org/majordomo-info.html
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=Y0TRxDRFRgFonhtH@coredump.intra.peff.net \
--to=peff@peff.net \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=jacob@initialcommit.io \
--cc=me@ttaylorr.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
Code repositories for project(s) associated with this public inbox
https://80x24.org/mirrors/git.git
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).