From: Junio C Hamano <gitster@pobox.com>
To: Usman Akinyemi <usmanakinyemi202@gmail.com>
Cc: git@vger.kernel.org, christian.couder@gmail.com,
johncai86@gmail.com, me@ttaylorr.com, ps@pks.im,
shejialuo@gmail.com, Christian Couder <chriscool@tuxfamily.org>
Subject: Re: [PATCH v3 8/8] builtin/checkout-index: stop using `the_repository`
Date: Fri, 07 Mar 2025 06:15:31 -0800 [thread overview]
Message-ID: <xmqqfrjpexks.fsf@gitster.g> (raw)
In-Reply-To: <CAPSxiM_fLNUvg86TtCHZTbsE_VnhKdMiDa3pfLPsLa3ThZfxmw@mail.gmail.com> (Usman Akinyemi's message of "Fri, 7 Mar 2025 06:45:29 +0530")
Usman Akinyemi <usmanakinyemi202@gmail.com> writes:
>> Hmph, if we are passing anything down to these code paths, I would
>> have expected that it would be an instance of "struct index_state".
>>
>> Do these two helper functions need anything other than that from the
>> repository instance?
> No, they do not. They could possibly do in the future and is there any
> reason why we might want to pass the "struct index_state" instead of
> the whole "struct repository" ?
You are asking a wrong question.
When there are two things, A and B, where B is a piece of data with
a smaller scope that is contained within A, unless your function
needs to access parts of A that is impossible to access if you fed
it only B, limiting the interface to the smallest piece necessary
(in this case, B) is always the better design. A potential caller
that only has B and not A can still call your function that takes B
if you designed the API that way. If you insist that the caller
pass A (and infer which B to use as part of A), a potential caller
that only has B cannot call your function.
So, with that understanding of a general principle in mind, you need
a stronger justification to pass the larger object A that goes
against the best practice, saying "Even though we do not have to
pass A because only B is necessary, we pass A because ...". And you
just said there is no such reason why you need a repository and for
these functions an index_state is sufficient.
There are many functions that only need "struct index_state"
instance and obtains one from their callers. They do not have a
repository object, other than the one that is referred to from the
index_state (to keep track of which repository it originated from).
But this pointer is not designed to round-trip. There can be, and
there indeed are, code paths that use multiple index_state instances
associated with one repository. But a repository instance has only
one ".index" member. It means if you pass a repository, you are
making it impossible for your potential callers that has secondary
index_state instances.
next prev parent reply other threads:[~2025-03-07 14:15 UTC|newest]
Thread overview: 63+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-02-14 22:57 [PATCH 0/7][Outreachy] stop using the_repository global variable Usman Akinyemi
2025-02-14 22:57 ` [PATCH 1/7] builtin/verify-tag: stop using `the_repository` Usman Akinyemi
2025-02-17 6:55 ` Patrick Steinhardt
2025-02-17 10:05 ` Usman Akinyemi
2025-02-17 10:22 ` Patrick Steinhardt
2025-02-17 10:42 ` Usman Akinyemi
2025-02-17 15:47 ` Patrick Steinhardt
2025-02-14 22:57 ` [PATCH 2/7] builtin/verify-commit.c: " Usman Akinyemi
2025-02-16 5:32 ` shejialuo
2025-02-17 8:55 ` Usman Akinyemi
2025-02-14 22:57 ` [PATCH 3/7] builtin/send-pack.c: " Usman Akinyemi
2025-02-14 22:57 ` [PATCH 4/7] builtin/pack-refs: " Usman Akinyemi
2025-02-14 22:57 ` [PATCH 5/7] builtin/ls-files: " Usman Akinyemi
2025-02-17 6:55 ` Patrick Steinhardt
2025-02-17 8:57 ` Usman Akinyemi
2025-02-14 22:57 ` [PATCH 6/7] builtin/for-each-ref: " Usman Akinyemi
2025-02-14 22:57 ` [PATCH 7/7] builtin/checkout-index.c: " Usman Akinyemi
2025-02-16 5:41 ` [PATCH 0/7][Outreachy] stop using the_repository global variable shejialuo
2025-02-17 8:56 ` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 00/12][Outreachy] " Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 01/12] builtin/verify-tag: refactor `cmd_verify_tag()` Usman Akinyemi
2025-02-20 15:32 ` Junio C Hamano
2025-02-19 20:32 ` [PATCH v2 02/12] builtin/verify-tag: stop using `the_repository` Usman Akinyemi
2025-02-20 15:43 ` Junio C Hamano
2025-02-27 17:56 ` Usman Akinyemi
2025-02-27 22:39 ` Junio C Hamano
2025-02-19 20:32 ` [PATCH v2 03/12] builtin/verify-commit: refactor `cmd_verify_commit()` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 04/12] builtin/verify-commit: stop using `the_repository` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 05/12] builtin/send-pack: refactor `cmd_send_pack()` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 06/12] builtin/send-pack: stop using `the_repository` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 07/12] builtin/pack-refs: refactor `cmd_pack_refs()` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 08/12] builtin/pack-refs: stop using `the_repository` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 09/12] builtin/ls-files: " Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 10/12] builtin/for-each-ref: refactor `cmd_for_each_ref()` Usman Akinyemi
2025-02-19 20:32 ` [PATCH v2 11/12] builtin/for-each-ref: stop using `the_repository` Usman Akinyemi
2025-02-19 20:33 ` [PATCH v2 12/12] builtin/checkout-index: " Usman Akinyemi
2025-03-06 14:35 ` [Outreachy PATCH v3 0/8] stop using the_repository global variable Usman Akinyemi
2025-03-06 14:35 ` [PATCH v3 1/8] config: teach repo_config to allow `repo` to be NULL Usman Akinyemi
2025-03-06 17:53 ` Junio C Hamano
2025-03-07 1:33 ` Usman Akinyemi
2025-03-07 10:37 ` Phillip Wood
2025-03-06 14:35 ` [PATCH v3 2/8] builtin/verify-tag: stop using `the_repository` Usman Akinyemi
2025-03-06 17:56 ` Junio C Hamano
2025-03-06 14:35 ` [PATCH v3 3/8] builtin/verify-commit: " Usman Akinyemi
2025-03-06 14:35 ` [PATCH v3 4/8] builtin/send-pack: " Usman Akinyemi
2025-03-06 14:35 ` [PATCH v3 5/8] builtin/pack-refs: " Usman Akinyemi
2025-03-06 14:35 ` [PATCH v3 6/8] builtin/ls-files: " Usman Akinyemi
2025-03-06 17:59 ` Junio C Hamano
2025-03-06 14:35 ` [PATCH v3 7/8] builtin/for-each-ref: " Usman Akinyemi
2025-03-06 14:35 ` [PATCH v3 8/8] builtin/checkout-index: " Usman Akinyemi
2025-03-06 18:18 ` Junio C Hamano
2025-03-07 1:15 ` Usman Akinyemi
2025-03-07 14:15 ` Junio C Hamano [this message]
2025-03-07 19:41 ` Usman Akinyemi
2025-03-07 23:34 ` [Outreachy PATCH v4 0/8] stop using the_repository global variable Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 1/8] config: teach repo_config to allow `repo` to be NULL Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 2/8] builtin/verify-tag: stop using `the_repository` Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 3/8] builtin/verify-commit: " Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 4/8] builtin/send-pack: " Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 5/8] builtin/pack-refs: " Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 6/8] builtin/ls-files: " Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 7/8] builtin/for-each-ref: " Usman Akinyemi
2025-03-07 23:35 ` [PATCH v4 8/8] builtin/checkout-index: " Usman Akinyemi
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=xmqqfrjpexks.fsf@gitster.g \
--to=gitster@pobox.com \
--cc=chriscool@tuxfamily.org \
--cc=christian.couder@gmail.com \
--cc=git@vger.kernel.org \
--cc=johncai86@gmail.com \
--cc=me@ttaylorr.com \
--cc=ps@pks.im \
--cc=shejialuo@gmail.com \
--cc=usmanakinyemi202@gmail.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.
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 NNTP newsgroup(s).