git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>
To: Glen Choo <chooglen@google.com>
Cc: git@vger.kernel.org
Subject: Re: [RFC PATCH 0/8] Get rid of "git --super-prefix"
Date: Tue, 15 Nov 2022 02:37:45 +0100	[thread overview]
Message-ID: <221115.864jv1kn4e.gmgdl@evledraar.gmail.com> (raw)
In-Reply-To: <kl6lmt8tnmbt.fsf@chooglen-macbookpro.roam.corp.google.com>


On Mon, Nov 14 2022, Glen Choo wrote:

> Ævar Arnfjörð Bjarmason <avarab@gmail.com> writes:
>
>> As noted in the CL above I included this because I see you're keen to
>> include it, but I'm personally a bit "meh" on it. I.e. it's just
>> renaming an existing unrelated option, although being able to use
>> OPT__SUPER_PREFIX() makes it slightly nicer.
>>
>> As post-cleanups go I think removing the "submodule_prefix" from the
>> "struct repository" would make more sense, and maybe it's worth peeling
>> off the 10/10 to include in such a post-cleanup series? I.e. the below
>> on top of all of this works, and reduces allocations and cargo-culting
>> around the submodule API.
>
> As a first impression I'm not particularly keen on this, since it makes
> perfect sense to me to have a repo->submodule_prefix, especially when
> recursing into N-level deep submodules...

This is mainly a quick experiment I drafted up, with that context...

>> -- >8 --
>> Subject: [PATCH] repo & submodule API: stop carrying "repo->submodule_prefix"
>>
>> As this change shows the "submodule_prefix" field to "struct
>> repository" added in 96dc883b3cd (repository: enable initialization of
>> submodules, 2017-06-22) was only used by "ls-files" and "grep". Let's
>> have those two carry forward the "super_prefix" instead.
>>
>> Having every user of "struct repository" needing to worry about this
>> created a mismatch in the API where e.g. "grep" would re-compute a
>> "name_base_len" which we knew before. Now we use a "struct strbuf" in
>> the "struct grep_opt" there instead, so we'll know the length
>> already. This simplifies "grep_cache()" and "grep_tree()".
>>
>> We're also deleting cargo-culted code that the previous API foisted
>> upon us. In 605f0ec1350 (submodule: use submodule repos for object
>> lookup, 2018-12-14) the "Mark it as a submodule" code was added to
>> "open_submodule()", but the resulting xstrdup()'d "submodule_prefix"
>> was never used by anything.
>
> (As an aside, I think open_submodule() should have been replaced by
> repo_submodule_init().)
>
> In which case, yes it isn't used by anything in that code path, but
> being meticulous about maintaining .super_prefix means that other
> callers could use it if they wanted to, which might be crucial once we
> start plumb "struct repository" deeper and deeper and...

...I just don't see the point of maintaining an API with this sort of
reach for just two callers, when it's a bad fit for what those two
callers need.

>>
>> Still, removing this field might not be a good move, as the
>> "super_prefix" might be a common enough case in the future, e.g. when
>> eventually migrating the "submodule--helper" users[1] to run
>> in-process.
>>
>> As the "grep" example demonstrates I don't think that's the
>> case. There instead of xstrdup()-ing all the way down we're now
>> carrying a single "super_prefix" in the form of a "struct strbuf". As
>> we recurse we then append to it, and strbuf_setlen() it back when we
>> we recurse out of that submodule. This is similar to how e.g. the
>> "read_tree_at()" API works.
>
> This technique might no longer be so appealing. We _could_ pass both
> "struct repository" and "super_prefix", but that seems odd given that
> the super prefix is tied to the repository.

FWIW I don't even think a "super_prefix" being tied to a repository is a
useful abstraction level in general.

E.g. I've got some local hacky patches to make "git ls-tree" able to
recurse down submodules, and for that code you usually just want to
print the prefix to the current *dir*, and having to print the prefix to
the current module, then the current dir-within-module, then the file
just makes things needlessly complex.

> But that's just a first impression anyway. I don't mind taking another
> look if this gets a standalone review.

*nod*

  reply	other threads:[~2022-11-15  1:44 UTC|newest]

Thread overview: 105+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-11-09  0:47 [RFC PATCH 0/4] git: remove --super-prefix Glen Choo
2022-11-09  0:47 ` [RFC PATCH 1/4] submodule--helper: teach --toplevel-cwd-prefix Glen Choo
2022-11-09  2:37   ` Ævar Arnfjörð Bjarmason
2022-11-09  0:47 ` [RFC PATCH 2/4] fetch: refactor --submodule-prefix Glen Choo
2022-11-09  3:06   ` Ævar Arnfjörð Bjarmason
2022-11-09  0:47 ` [RFC PATCH 3/4] read-tree: teach --submodule-prefix Glen Choo
2022-11-09  3:13   ` Ævar Arnfjörð Bjarmason
2022-11-09  0:47 ` [RFC PATCH 4/4] git: remove --super-prefix Glen Choo
2022-11-09 19:34 ` [RFC PATCH 0/8] Get rid of "git --super-prefix" Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 1/8] submodule--helper: don't use global --super-prefix in "absorbgitdirs" Ævar Arnfjörð Bjarmason
2022-11-11  0:12     ` Glen Choo
2022-11-09 19:34   ` [RFC PATCH 2/8] submodule--helper: "deinit" has never used "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 3/8] submodule--helper: convert "foreach" to its own "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 4/8] submodule--helper: convert "sync" " Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 5/8] submodule--helper: convert "status" " Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 6/8] submodule--helper: convert "{update,clone}" to their " Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 7/8] submodule tests: test "git branch -t" output and stderr Ævar Arnfjörð Bjarmason
2022-11-09 19:34   ` [RFC PATCH 8/8] read-tree: add "--super-prefix" option, eliminate global Ævar Arnfjörð Bjarmason
2022-11-11  0:40     ` Glen Choo
2022-11-09 21:21   ` [RFC PATCH 0/8] Get rid of "git --super-prefix" Taylor Blau
2022-11-09 21:47     ` Ævar Arnfjörð Bjarmason
2022-11-09 22:27       ` Taylor Blau
2022-11-09 22:54         ` Ævar Arnfjörð Bjarmason
2022-11-10  0:45   ` Glen Choo
2022-11-10 10:51     ` Ævar Arnfjörð Bjarmason
2022-11-11  1:07       ` Glen Choo
2022-11-11 18:29         ` Glen Choo
2022-11-11 21:17           ` Ævar Arnfjörð Bjarmason
2022-11-11 21:51             ` Taylor Blau
2022-11-12  1:10             ` Glen Choo
2022-11-14 10:09               ` Ævar Arnfjörð Bjarmason
2022-11-14 23:33                 ` Glen Choo
2022-11-15  1:37                   ` Ævar Arnfjörð Bjarmason [this message]
2022-11-14 10:08   ` [PATCH v2 00/10] " Ævar Arnfjörð Bjarmason
2022-11-14 10:08     ` [PATCH v2 01/10] read-tree + fetch tests: test failing "--super-prefix" interaction Ævar Arnfjörð Bjarmason
2022-11-14 19:00       ` Glen Choo
2022-11-14 19:14         ` Ævar Arnfjörð Bjarmason
2022-11-14 19:49           ` Glen Choo
2022-11-14 10:08     ` [PATCH v2 02/10] submodule--helper: don't use global --super-prefix in "absorbgitdirs" Ævar Arnfjörð Bjarmason
2022-11-14 21:22       ` Glen Choo
2022-11-17 18:10         ` Ævar Arnfjörð Bjarmason
2022-11-14 10:08     ` [PATCH v2 03/10] submodule--helper: "deinit" has never used "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-14 10:08     ` [PATCH v2 04/10] submodule--helper: convert "foreach" to its own "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-14 21:56       ` Glen Choo
2022-11-17 18:14         ` Ævar Arnfjörð Bjarmason
2022-11-14 10:08     ` [PATCH v2 05/10] submodule--helper: convert "sync" " Ævar Arnfjörð Bjarmason
2022-11-14 10:08     ` [PATCH v2 06/10] submodule--helper: convert "status" " Ævar Arnfjörð Bjarmason
2022-11-14 10:08     ` [PATCH v2 07/10] submodule--helper: convert "{update,clone}" to their " Ævar Arnfjörð Bjarmason
2022-11-14 22:04       ` Glen Choo
2022-11-14 10:08     ` [PATCH v2 08/10] submodule tests: test "git branch -t" output and stderr Ævar Arnfjörð Bjarmason
2022-11-14 22:20       ` Glen Choo
2022-11-14 10:08     ` [PATCH v2 09/10] read-tree: add "--super-prefix" option, eliminate global Ævar Arnfjörð Bjarmason
2022-11-14 22:28       ` Glen Choo
2022-11-14 10:08     ` [PATCH v2 10/10] fetch: rename "--submodule-prefix" to "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-14 22:31       ` Glen Choo
2022-11-14 21:59     ` [PATCH v2 00/10] Get rid of "git --super-prefix" Taylor Blau
2022-11-14 23:20     ` Glen Choo
2022-11-14 23:39     ` Glen Choo
2022-11-15  1:27       ` Ævar Arnfjörð Bjarmason
2022-11-16 21:07         ` Glen Choo
2022-11-17 18:07           ` Ævar Arnfjörð Bjarmason
2022-11-21 19:16             ` Glen Choo
2022-11-19 12:41     ` [PATCH v3 0/9] " Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 1/9] read-tree + fetch tests: test failing "--super-prefix" interaction Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 2/9] submodule.c & submodule--helper: pass along "super_prefix" param Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 3/9] submodule--helper: don't use global --super-prefix in "absorbgitdirs" Ævar Arnfjörð Bjarmason
2022-11-22 19:53         ` Glen Choo
2022-11-19 12:41       ` [PATCH v3 4/9] submodule--helper: convert "foreach" to its own "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 5/9] submodule--helper: convert "sync" " Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 6/9] submodule--helper: convert "status" " Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 7/9] submodule--helper: convert "{update,clone}" to their " Ævar Arnfjörð Bjarmason
2022-11-19 12:41       ` [PATCH v3 8/9] read-tree: add "--super-prefix" option, eliminate global Ævar Arnfjörð Bjarmason
2022-11-22 19:57         ` Glen Choo
2022-11-19 12:41       ` [PATCH v3 9/9] fetch: rename "--submodule-prefix" to "--super-prefix" Ævar Arnfjörð Bjarmason
2022-11-22 22:29       ` [PATCH v3 0/9] Get rid of "git --super-prefix" Glen Choo
2022-12-15  9:32       ` [PATCH v4 " Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 1/9] submodule absorbgitdirs tests: add missing "Migrating git..." tests Ævar Arnfjörð Bjarmason
2022-12-15 20:54           ` Glen Choo
2022-12-20 10:32             ` Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 2/9] read-tree + fetch tests: test failing "--super-prefix" interaction Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 3/9] submodule.c & submodule--helper: pass along "super_prefix" param Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 4/9] submodule--helper: don't use global --super-prefix in "absorbgitdirs" Ævar Arnfjörð Bjarmason
2022-12-15 21:05           ` Glen Choo
2022-12-15  9:32         ` [PATCH v4 5/9] submodule--helper: convert "foreach" to its own "--super-prefix" Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 6/9] submodule--helper: convert "sync" " Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 7/9] submodule--helper: convert "status" " Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 8/9] submodule--helper: convert "{update,clone}" to their " Ævar Arnfjörð Bjarmason
2022-12-15  9:32         ` [PATCH v4 9/9] read-tree: add "--super-prefix" option, eliminate global Ævar Arnfjörð Bjarmason
2022-12-15 21:19         ` [PATCH v4 0/9] Get rid of "git --super-prefix" Glen Choo
2022-12-15 22:19           ` Junio C Hamano
2022-12-15 22:12         ` Junio C Hamano
2022-12-20 12:39         ` [PATCH v5 " Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 1/9] submodule absorbgitdirs tests: add missing "Migrating git..." tests Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 2/9] read-tree + fetch tests: test failing "--super-prefix" interaction Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 3/9] submodule.c & submodule--helper: pass along "super_prefix" param Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 4/9] submodule--helper: don't use global --super-prefix in "absorbgitdirs" Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 5/9] submodule--helper: convert "foreach" to its own "--super-prefix" Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 6/9] submodule--helper: convert "sync" " Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 7/9] submodule--helper: convert "status" " Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 8/9] submodule--helper: convert "{update,clone}" to their " Ævar Arnfjörð Bjarmason
2022-12-20 12:39           ` [PATCH v5 9/9] read-tree: add "--super-prefix" option, eliminate global Ævar Arnfjörð Bjarmason
2022-11-09 21:16 ` [RFC PATCH 0/4] git: remove --super-prefix Taylor Blau
2022-11-09 23:55   ` Glen Choo
2022-11-10  2:14     ` Taylor Blau
2022-11-10 23:49       ` Glen Choo

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=221115.864jv1kn4e.gmgdl@evledraar.gmail.com \
    --to=avarab@gmail.com \
    --cc=chooglen@google.com \
    --cc=git@vger.kernel.org \
    /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).