From: Jeff King <peff@peff.net>
To: Brandon Williams <bmwill@google.com>
Cc: Jonathan Nieder <jrnieder@gmail.com>,
git@vger.kernel.org, gitster@pobox.com
Subject: Re: [PATCH 4/4] config: don't implicitly use gitdir
Date: Tue, 13 Jun 2017 02:45:32 -0400 [thread overview]
Message-ID: <20170613064532.pucjnyrftulmjinj@sigill.intra.peff.net> (raw)
In-Reply-To: <20170613061627.GJ154599@google.com>
On Mon, Jun 12, 2017 at 11:16:27PM -0700, Brandon Williams wrote:
> > If the parameter is now required, then it might make sense for it to
> > become an actual function parameter instead of being stuffed into the
> > config_options struct. That would give you your breaking change, plus
> > make it more obvious to the reader that it is not optional.
> >
> > The downside is that has to get shuttled around manually through the
> > callstack. Most of the damage is in builtin/config.c, where we call
> > git_config_with_options() a lot.
> >
> > include_by_gitdir is also a bit annoying, as we pass around the
> > config_options struct through our void-pointer callbacks. But we can
> > solve that by sticking the git_dir into the include_data struct (whose
> > exact purpose is to carry the information we need to handle includes).
> >
> > The patch below (on top of Brandon's series does that).
>
> I really don't understand why this has to be so difficult and why a
> 'breaking change' is even needed. Duy just added the 'git_dir' field to
> the config_options struct in April of this year (2185fde56 config:
> handle conditional include when $GIT_DIR is not set up) and now we want
> to strip it out again? That's not even two months. Seems very counter
> productive and makes the api more unwieldy.
I could go either way on it. But note that you're not just changing the
existing opt->git_dir behavior.
If I call git_config_with_options() without having set opt->git_dir, the
call will now quietly ignore repo config. But even before opt->git_dir
existed, calling that function would always have read from repo config
(when we're in one, of course). So if there's a patch in flight that
adds a call to git_config_with_options(), it's now very subtly broken.
The reason I say "I could go either way" is that we can make a guess as
to whether there are any topics in flight that add such a call.
There aren't any in pu right now. That's not the whole world, of course;
people may have topics they haven't yet published. Or they may have long
running forks. Git for Windows is one, and I maintain one that GitHub
uses internally. But GfW is public and doesn't have any new calls (and
nor does my fork). In general, it's kind of an unlikely call for a fork
or a new branch to add.
So at some point I think we say "good enough, it's not worth the hassle"
and this may be such a case.
-Peff
next prev parent reply other threads:[~2017-06-13 6:45 UTC|newest]
Thread overview: 57+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-06-12 21:34 [PATCH 0/4] config.h Brandon Williams
2017-06-12 21:34 ` [PATCH 1/4] config: create config.h Brandon Williams
2017-06-12 21:34 ` [PATCH 2/4] config: remove git_config_iter Brandon Williams
2017-06-13 0:49 ` Jonathan Nieder
2017-06-13 0:57 ` Jeff King
2017-06-12 21:34 ` [PATCH 3/4] config: don't include config.h by default Brandon Williams
2017-06-12 21:34 ` [PATCH 4/4] config: don't implicitly use gitdir Brandon Williams
2017-06-13 1:05 ` Jonathan Nieder
2017-06-13 1:23 ` Brandon Williams
2017-06-13 1:33 ` Jonathan Nieder
2017-06-13 1:38 ` Jonathan Nieder
2017-06-13 2:59 ` Jeff King
2017-06-13 6:16 ` Brandon Williams
2017-06-13 6:45 ` Jeff King [this message]
2017-06-13 7:08 ` Jeff King
2017-06-13 14:43 ` Brandon Williams
2017-06-13 17:06 ` Jonathan Nieder
2017-06-13 5:52 ` Brandon Williams
2017-06-13 6:29 ` Jeff King
2017-06-13 14:47 ` Brandon Williams
2017-06-12 21:45 ` [PATCH 0/4] config.h Jeff King
2017-06-12 21:53 ` Brandon Williams
2017-06-12 22:02 ` Jeff King
2017-06-12 22:06 ` Brandon Williams
2017-06-13 1:07 ` Jonathan Nieder
2017-06-13 21:03 ` [PATCH v2 0/6] config.h Brandon Williams
2017-06-13 21:03 ` [PATCH v2 1/6] config: create config.h Brandon Williams
2017-06-13 21:13 ` Jonathan Nieder
2017-06-13 21:03 ` [PATCH v2 2/6] config: remove git_config_iter Brandon Williams
2017-06-13 21:14 ` Jonathan Nieder
2017-06-13 21:03 ` [PATCH v2 3/6] config: don't include config.h by default Brandon Williams
2017-06-13 21:58 ` Jonathan Nieder
2017-06-13 21:03 ` [PATCH v2 4/6] config: don't implicitly use gitdir Brandon Williams
2017-06-13 21:08 ` Jonathan Nieder
2017-06-13 21:38 ` Brandon Williams
2017-06-13 21:51 ` Jonathan Nieder
2017-06-13 21:55 ` Junio C Hamano
2017-06-13 22:05 ` Jonathan Nieder
2017-06-14 4:40 ` Jacob Keller
2017-06-14 6:25 ` Jeff King
2017-06-14 17:14 ` Brandon Williams
2017-06-13 21:03 ` [PATCH v2 5/6] setup: teach discover_git_directory to respect the commondir Brandon Williams
2017-06-14 6:15 ` Jeff King
2017-06-14 17:19 ` Brandon Williams
2017-06-13 21:03 ` [PATCH v2 6/6] config: respect commondir Brandon Williams
2017-06-14 18:07 ` [PATCH v3 0/6] config.h Brandon Williams
2017-06-14 18:07 ` [PATCH v3 1/6] config: create config.h Brandon Williams
2017-06-14 18:07 ` [PATCH v3 2/6] config: remove git_config_iter Brandon Williams
2017-06-14 18:07 ` [PATCH v3 3/6] config: don't include config.h by default Brandon Williams
2017-06-14 18:07 ` [PATCH v3 4/6] setup: teach discover_git_directory to respect the commondir Brandon Williams
2017-06-14 18:07 ` [PATCH v3 5/6] config: respect commondir Brandon Williams
2017-06-14 18:07 ` [PATCH v3 6/6] config: don't implicitly use gitdir or commondir Brandon Williams
2017-06-15 19:59 ` [PATCH v3 0/6] config.h Junio C Hamano
2017-06-15 20:33 ` Brandon Williams
2017-06-15 21:09 ` Junio C Hamano
2017-06-15 21:18 ` Brandon Williams
2017-06-16 0:12 ` Brandon Williams
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=20170613064532.pucjnyrftulmjinj@sigill.intra.peff.net \
--to=peff@peff.net \
--cc=bmwill@google.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=jrnieder@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.
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).