git@vger.kernel.org mailing list mirror (one of many)
 help / color / mirror / code / Atom feed
From: Duy Nguyen <pclouds@gmail.com>
To: Jeff King <peff@peff.net>
Cc: Rafael Ascensao <rafa.almas@gmail.com>,
	Git Mailing List <git@vger.kernel.org>
Subject: Re: git complains packed-refs is not a directory when used with GIT_DIR and GIT_WORK_TREE envvars.
Date: Tue, 27 Mar 2018 19:09:36 +0200	[thread overview]
Message-ID: <CACsJy8Bog6U7X-jvzDhq14heQWx0HA_21HsSYR0nykU9aDsCYQ@mail.gmail.com> (raw)
In-Reply-To: <20180327164757.GB24747@sigill.intra.peff.net>

On Tue, Mar 27, 2018 at 6:47 PM, Jeff King <peff@peff.net> wrote:
>> So I would not mind papering over it right now (with an understanding
>> that absolute path pays some more overhead in path walking, which was
>> th reason we tried to avoid it in setup code). A slightly better patch
>> is trigger this path absolutization from setup_work_tree(), near
>> set_git_dir(). But then you face some ugliness there: how to find out
>> all ref stores to update, or just update the main ref store alone.
>
> I don't quite get why f57f37e2 doesn't want to call git_path(). Is it to
> avoid the way the path is munged? Or is it to avoid some lazy-setup that
> is triggered by calling get_git_dir() at all (which doesn't make much
> sense to me, because we'd already have called get_git_dir() much
> earlier). Or is it just because we may sometimes fill in refs->git_dir
> with something besides get_git_dir() (for a submodule or worktree or
> something)?

None of those, I think. git_path() does some magic to translate paths
so that refs/... ends up with $GIT_COMMON_DIR/refs/... while "index"
ends up with $GIT_DIR/index. Michael wanted to avoid that magic and
keep the control within refs code (i.e. this code knows refs/ and
packed-refs are shared, and pseudo refs are not, what git_path()
decides does not matter).

> I.e., can we do one of (depending on which of those answers is "yes"):
>
>   1. Stop caching the value of get_git_dir(), and instead call it
>      on-demand instead of looking at refs->git_dir? (If we just want to
>      avoid git_path()).

This probably works, but I count it as papering over the problem too.

>
>   2. If we need to avoid even calling get_git_dir(), can we add a
>      "light" version of it that avoids whatever side effects we're
>      trying to skip?
>
>   3. If the problem is just that sometimes we need get_git_dir() and
>      sometimes not, could we perhaps store NULL as a sentinel to mean
>      "look up get_git_dir() when you need it"?
>
>      That would let submodules and worktrees fill in their paths as
>      necessary (assuming they never change after init), but handle the
>      case of get_git_dir() changing.
>
> Hmm. Typing that out, it seems like (3) is probably the right path.
> Something like the patch below seems to fix it and passes the tests.

Honestly I think this is just another way to work around the problem
(with even more changes than your abspath approach). The problem is
with setup_work_tree(). We create a ref store at a specific location
and it should stay working without lazily calling get_git_dir(), which
has nothing to do (anymore) with the path we have given a ref store.
If somebody changes a global setting like $CWD, it should be well
communicated to everybody involved.

I would rather have something like ref_store_reinit() in the same
spirit as the second call of set_git_dir() in setup_work_tree. It is
hacky, but it works and keeps changes to minimal (so that it could be
easily replaced later).
-- 
Duy

  reply	other threads:[~2018-03-27 17:10 UTC|newest]

Thread overview: 54+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2018-03-26 21:27 git complains packed-refs is not a directory when used with GIT_DIR and GIT_WORK_TREE envvars Rafael Ascensao
2018-03-26 21:44 ` Ævar Arnfjörð Bjarmason
2018-03-27  6:31 ` Jeff King
2018-03-27 14:56   ` Duy Nguyen
2018-03-27 16:47     ` Jeff King
2018-03-27 17:09       ` Duy Nguyen [this message]
2018-03-27 17:30         ` Duy Nguyen
2018-03-28  9:52           ` Jeff King
2018-03-28 10:10             ` Duy Nguyen
2018-03-28 17:36               ` Jeff King
2018-03-28 17:38                 ` [PATCH 1/4] set_git_dir: die when setenv() fails Jeff King
2018-03-28 17:40                 ` [PATCH 2/4] add chdir-notify API Jeff King
2018-03-28 17:58                   ` Eric Sunshine
2018-03-28 18:02                     ` Jeff King
2018-03-29 14:53                   ` Duy Nguyen
2018-03-29 17:48                     ` Jeff King
2018-03-29 18:12                       ` Duy Nguyen
2018-03-28 17:42                 ` [PATCH 3/4] set_work_tree: use chdir_notify Jeff King
2018-03-29 17:02                   ` Duy Nguyen
2018-03-29 17:23                     ` Duy Nguyen
2018-03-29 17:50                       ` Jeff King
2018-03-29 17:50                     ` Jeff King
2018-03-29 18:01                       ` Duy Nguyen
2018-03-30 17:23                         ` Jeff King
2018-03-28 17:43                 ` [PATCH 4/4] refs: use chdir_notify to update cached relative paths Jeff King
2018-03-30 18:34                 ` [PATCH v2 0/5] re-parenting relative directories after chdir Jeff King
2018-03-30 18:34                   ` [PATCH v2 1/5] set_git_dir: die when setenv() fails Jeff King
2018-03-30 18:34                   ` [PATCH v2 2/5] trace.c: export trace_setup_key Jeff King
2018-03-30 19:46                     ` Junio C Hamano
2018-03-30 19:47                       ` Jeff King
2018-03-30 19:50                         ` Junio C Hamano
2018-03-30 19:54                           ` Jeff King
2018-03-30 18:35                   ` [PATCH v2 3/5] add chdir-notify API Jeff King
2018-03-30 18:35                   ` [PATCH v2 4/5] set_work_tree: use chdir_notify Jeff King
2018-03-30 18:35                   ` [PATCH v2 5/5] refs: use chdir_notify to update cached relative paths Jeff King
2018-03-30 19:36                   ` [PATCH v2 0/5] re-parenting relative directories after chdir Duy Nguyen
2018-03-28  9:47         ` git complains packed-refs is not a directory when used with GIT_DIR and GIT_WORK_TREE envvars Jeff King
2018-03-28 17:55           ` [PATCH 0/8] " Nguyễn Thái Ngọc Duy
2018-03-28 17:55             ` [PATCH 1/8] strbuf.c: add strbuf_ensure_trailing_dr_sep() Nguyễn Thái Ngọc Duy
2018-03-28 17:55             ` [PATCH 2/8] strbuf.c: reintroduce get_pwd_cwd() (with strbuf_ prefix) Nguyễn Thái Ngọc Duy
2018-03-28 18:02               ` Stefan Beller
2018-03-28 18:05                 ` Duy Nguyen
2018-03-28 17:55             ` [PATCH 3/8] trace.c: export trace_setup_key Nguyễn Thái Ngọc Duy
2018-03-28 17:55             ` [PATCH 4/8] setup.c: introduce setup_adjust_path() Nguyễn Thái Ngọc Duy
2018-03-28 17:55             ` [PATCH 5/8] setup.c: allow other code to be notified when $CWD moves Nguyễn Thái Ngọc Duy
2018-03-28 17:55             ` [PATCH 6/8] environment.c: adjust env containing relpath when $CWD is moved Nguyễn Thái Ngọc Duy
2018-03-28 18:30               ` Jeff King
2018-03-28 18:45                 ` Duy Nguyen
2018-03-28 17:55             ` [PATCH 7/8] repository: adjust repo paths when $CWD moves Nguyễn Thái Ngọc Duy
2018-03-28 17:55             ` [PATCH 8/8] refs: adjust main " Nguyễn Thái Ngọc Duy
2018-03-28 18:19             ` [PATCH 0/8] Re: git complains packed-refs is not a directory when used with GIT_DIR and GIT_WORK_TREE envvars Jeff King
2018-03-29 14:57               ` Duy Nguyen
2018-03-30 17:21                 ` Jeff King
2018-03-28 22:24             ` Junio C Hamano

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=CACsJy8Bog6U7X-jvzDhq14heQWx0HA_21HsSYR0nykU9aDsCYQ@mail.gmail.com \
    --to=pclouds@gmail.com \
    --cc=git@vger.kernel.org \
    --cc=peff@peff.net \
    --cc=rafa.almas@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).