From: Victoria Dye <vdye@github.com>
To: Derrick Stolee <derrickstolee@github.com>,
Matthew John Cheetham via GitGitGadget <gitgitgadget@gmail.com>,
git@vger.kernel.org
Cc: johannes.schindelin@gmx.de, mjcheetham@outlook.com, gitster@pobox.com
Subject: Re: [PATCH v2 3/5] scalar: enable built-in FSMonitor on `register`
Date: Wed, 17 Aug 2022 16:47:13 -0700 [thread overview]
Message-ID: <3a238691-c37f-39a1-f4f6-2b8f9b4c4dcb@github.com> (raw)
In-Reply-To: <f5388e4d-7eb7-9333-6a8e-86ce449aced0@github.com>
Derrick Stolee wrote:
> On 8/16/2022 7:58 PM, Matthew John Cheetham via GitGitGadget wrote:
>
>> +#ifdef HAVE_FSMONITOR_DAEMON_BACKEND
>> + /*
>> + * Enable the built-in FSMonitor on supported platforms.
>> + */
>> + { "core.fsmonitor", "true" },
>> +#endif
>> + if (fsmonitor_ipc__is_supported() && start_fsmonitor_daemon())
>> + return error(_("could not start the FSMonitor daemon"));
>> +
>
> I initially worried if fsmonitor_ipc__is_supported() could use some
> run-time information to detect if FS Monitor is supported (say, existence
> of a network share or something). However, that implementation is
> currently defined as a constant depending on
> HAVE_FSMONITOR_DAEMON_BACKEND.
>
> The reason I was worried is that we could enable core.fsmonitor=true based
> on the compile-time macro, but then avoid starting the daemon based on the
> run-time results. If we get into this state, would the user's 'git status'
> calls start complaining about the core.fsmonitor=true config because it is
> not supported?
>
> The most future-proof thing to do might be to move the config write out of
> the set_recommended_config() and into start_fsmonitor_daemon(). Perhaps
> rename it to enable_fsmonitor() so it can fail due to writing the config
> _or_ for starting the daemon. The error message would change, then, too.
I spent some time digging into this, and I think gating both the config and
subsequent 'git fsmonitor--daemon start' on having platform *and* repository
support is a good idea. I'll update the next version to both set the
'core.fsmonitor' config and start the daemon only if the built-in FSMonitor
is fully supported.
(warning: long-winded tangent mostly unrelated to FSMonitor)
In the process of testing FSMonitor behavior, I think found other issues
with Scalar registration. Specifically, the test I wrote attempted to
'scalar register' a bare repo, since bare directories are incompatible with
FSMonitor. After seeing that FSMonitor was *not* incompatible with the
repository, I found that Scalar was 1) ignoring the bare repository and, as
a result, 2) identifying my Git clone (way above GIT_CEILING_DIRECTORIES) as
the "enlistment root". I think 1) might be fine as-is - uniformly ignoring
bare repos seems like a reasonable choice - but 2) seems like more of a
problem.
Right now, 'setup_enlistment_directory()' searches for the repo root
beginning at directory '<dir>', which is either a user-provided path or
current working directory. It checks whether '<dir>' or '<dir>/src' is a
repo root: if so, it sets the enlistment info; otherwise, it repeats the
process with the parent of '<dir>' until the repo root is found. For
example, given the following directory structure:
somedir
└── enlistment
├── src
│ └── .git
└── test
└── data
'scalar register somedir/enlistment/test/data' will search:
* somedir/enlistment/test/data/src
* somedir/enlistment/test/data
* somedir/enlistment/test/src
* somedir/enlistment/test
* somedir/enlistment/src
The current usage of GIT_CEILING_DIRECTORIES relies on the fact that, when
invoking a normal 'git' command, 'setup_git_directory()' only searches
upwards from the current working directory to find the repo root; it's a
clear "yes" or "no" as to whether that search passes a ceiling directory.
Scalar isn't as clear, since it searches for the repo root both "downwards"
into '<dir>/src' *and* upwards through the parents of '<dir>'. It's not
totally clear to me what the "right" behavior for Scalar is, but my current
thought is to follow the same rules as 'setup_git_directory()', but for the
*enlistment* root rather than the repository root. It's more restrictive
than GIT_CEILING_DIRECTORIES on a normal git repo, e.g.:
1. 'GIT_CEILING_DIRECTORIES=somedir/enlistment git -C somedir/enlistment/src status'
is valid.
2. 'GIT_CEILING_DIRECTORIES=somedir/enlistment scalar register somedir/enlistment/src'
is not valid.
but since Scalar works on the entire enlistment (not just the repo inside of
it), I think it makes sense to prevent it from crossing a ceiling directory
boundary.
What do you think? Hopefully my rambling wasn't too confusing (if it is,
please let me know what I can clarify).
>
> Or maybe I'm making a mountain out of a mole hill and what exists here is
> perfectly fine.
>
>> +test_lazy_prereq BUILTIN_FSMONITOR '
>> + git version --build-options | grep -q "feature:.*fsmonitor--daemon"
>> +'
>
> It looks like we already have a FSMONITOR_DAEMON prereq in test-lib.sh.
> Should we use that instead?
Works for me, happy to reuse code wherever possible. :)
>
> Thanks,
> -Stolee
next prev parent reply other threads:[~2022-08-17 23:48 UTC|newest]
Thread overview: 39+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-16 18:07 [PATCH 0/3] scalar: enable built-in FSMonitor Victoria Dye via GitGitGadget
2022-08-16 18:07 ` [PATCH 1/3] scalar: enable built-in FSMonitor on `register` Matthew John Cheetham via GitGitGadget
2022-08-16 20:49 ` Junio C Hamano
2022-08-16 21:57 ` Victoria Dye
2022-08-16 22:15 ` Junio C Hamano
2022-08-16 18:07 ` [PATCH 2/3] scalar unregister: stop FSMonitor daemon Johannes Schindelin via GitGitGadget
2022-08-16 18:07 ` [PATCH 3/3] scalar: update technical doc roadmap with FSMonitor support Victoria Dye via GitGitGadget
2022-08-16 18:21 ` [PATCH 0/3] scalar: enable built-in FSMonitor Junio C Hamano
2022-08-16 18:42 ` Victoria Dye
2022-08-16 18:44 ` Junio C Hamano
2022-08-16 23:58 ` [PATCH v2 0/5] " Victoria Dye via GitGitGadget
2022-08-16 23:58 ` [PATCH v2 1/5] scalar-unregister: handle error codes greater than 0 Victoria Dye via GitGitGadget
2022-08-17 14:33 ` Junio C Hamano
2022-08-16 23:58 ` [PATCH v2 2/5] scalar-[un]register: clearly indicate source of error Victoria Dye via GitGitGadget
2022-08-16 23:58 ` [PATCH v2 3/5] scalar: enable built-in FSMonitor on `register` Matthew John Cheetham via GitGitGadget
2022-08-17 14:34 ` Derrick Stolee
2022-08-17 15:54 ` Junio C Hamano
2022-08-17 23:47 ` Victoria Dye [this message]
2022-08-18 13:19 ` Derrick Stolee
2022-08-17 14:43 ` Junio C Hamano
2022-08-16 23:58 ` [PATCH v2 4/5] scalar unregister: stop FSMonitor daemon Johannes Schindelin via GitGitGadget
2022-08-17 14:39 ` Derrick Stolee
2022-08-17 17:36 ` Victoria Dye
2022-08-17 17:45 ` Derrick Stolee
2022-08-16 23:58 ` [PATCH v2 5/5] scalar: update technical doc roadmap with FSMonitor support Victoria Dye via GitGitGadget
2022-08-17 14:51 ` [PATCH v2 0/5] scalar: enable built-in FSMonitor Derrick Stolee
2022-08-18 21:40 ` [PATCH v3 0/8] " Victoria Dye via GitGitGadget
2022-08-18 21:40 ` [PATCH v3 1/8] scalar: constrain enlistment search Victoria Dye via GitGitGadget
2022-08-19 18:32 ` Derrick Stolee
2022-08-18 21:40 ` [PATCH v3 2/8] scalar-unregister: handle error codes greater than 0 Victoria Dye via GitGitGadget
2022-08-18 21:40 ` [PATCH v3 3/8] scalar-[un]register: clearly indicate source of error Victoria Dye via GitGitGadget
2022-08-18 21:40 ` [PATCH v3 4/8] scalar-delete: do not 'die()' in 'delete_enlistment()' Victoria Dye via GitGitGadget
2022-08-18 21:40 ` [PATCH v3 5/8] scalar: move config setting logic into its own function Victoria Dye via GitGitGadget
2022-08-18 21:40 ` [PATCH v3 6/8] scalar: enable built-in FSMonitor on `register` Matthew John Cheetham via GitGitGadget
2022-08-19 18:44 ` Derrick Stolee
2022-08-18 21:40 ` [PATCH v3 7/8] scalar unregister: stop FSMonitor daemon Johannes Schindelin via GitGitGadget
2022-08-18 21:40 ` [PATCH v3 8/8] scalar: update technical doc roadmap with FSMonitor support Victoria Dye via GitGitGadget
2022-08-19 18:45 ` [PATCH v3 0/8] scalar: enable built-in FSMonitor Derrick Stolee
2022-08-19 21:06 ` 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=3a238691-c37f-39a1-f4f6-2b8f9b4c4dcb@github.com \
--to=vdye@github.com \
--cc=derrickstolee@github.com \
--cc=git@vger.kernel.org \
--cc=gitgitgadget@gmail.com \
--cc=gitster@pobox.com \
--cc=johannes.schindelin@gmx.de \
--cc=mjcheetham@outlook.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).