Update May 2018: git help -c
, as noted in Petr's answer.
With Git 2.19 (Q2 2018), continuing with the idea to programatically enumerate various pieces of data required for command line completion, teach the codebase to report the list of configuration variables subcommands care about to help complete them.
See commit f22f682 (27 May 2018), and commit 09c4ba4, commit bea2125, commit f45db83, commit e17ca92, commit 431bb23, commit fb6fbff, commit a4a9cc1, commit 3ac68a9, commit a46baac, commit fa151dc, commit a73b368 (26 May 2018) by Nguyễn Thái Ngọc Duy (pclouds
).
See commit 17b3e51 (29 May 2018) by Junio C Hamano (gitster
).
(Merged by Junio C Hamano -- gitster
-- in commit ebaf0a5, 25 Jun 2018)
help
: add --config
to list all available config
Signed-off-by: Nguyễn Thái Ngọc Duy
Sometimes it helps to list all available config vars so the user can search for something they want. The config man page can also be used but it's harder to search if you want to focus on the variable name, for example.
This is not the best way to collect the available config since it's not precise. Ideally we should have a centralized list of config in C code (pretty much like 'struct option'), but that's a lot more work. This will do for now.
Strangely enough, this is formelly documented only with Git 2.34 (Q4 2021).
See commit ae578de (13 Sep 2021) by Philip Oakley (PhilipOakley
).
(Merged by Junio C Hamano -- gitster
-- in commit 68658a8, 23 Sep 2021)
doc
: config, tell readers of git help --config
Signed-off-by: Philip Oakley
git config
now includes in its man page:
A list of all available configuration variables can be obtained using the
git help --config
command.
Original answer: 2015
That was debated in this thread in 2013, requested by Sebastian Schuberth, Jeff King (Peff
) adding:
The expected output is certainly a problem, but the issue is more
fundamental than that: git config
does not even know what the default
is for any given option.
It is assumed that the caller knows what to do with an unset value. And
this is nothing to do with git config
; the internal C code works the
same way.
The actual defaults are not even necessarily expressible through the config.
E.g., I know that http.receivepack
considers "unset"
to be distinct either "true
" or "false
", but setting it can yield only
one of those latter two values.
I'm sure there are others, too (I just happened to notice that one this week).
(For instance: gc.prune
)
I could certainly see an argument that the world would be a better place
if the code had a big table of options and their descriptions, possible
values, and defaults, and if we used that to generate documentation as
well as validate input.
But nobody has gone to the trouble to construct that table and convert all of the callers. And as Jakub (Jakub Narębski) mentioned, such a central table can do nothing for external programs that store their config alongside git's.
In short:
git config
does not even know any of the options or values it manages,
but just is a "dumb" front-end to writing / reading whatever you pass
it to / from a file.
Note: git config was introduced in commit 1771299 (git 0.99.9a, Oct. 2005)
Different programs can react to different config options, although they
should always fall back to calling "git_default_config()" on any config
option name that they don't recognize.
So internally, there is a way to load default config, used as recently as commit 72549df, git 2.2.0-rc1, Nov. 2014, by the same Peff:
When we start the git-fetch program, we call git_config to load all config, but our callback only processes the fetch.prune
option; we do not chain to git_default_config
at all.
This means that we may not load some core configuration which will have an effect. For instance, we do not load core.logAllRefUpdates
, which impacts whether or not we create reflogs in a bare repository.
Let's just load the core config at the start of fetch, so we know we have it
See another example with commit 3e1dd17, git 1.7.7-rc1, Aug. 2011 with the default color config.
Completion can help too seeing a list of all config variables in git :
With Git 2.34 (Q4 2021), teach "git help -c
"(man) into helping the command line completion of configuration variables.
See commit 06fa4db, commit a9baccc, commit 5a5f04d, commit d35d03c, commit 0a5940f, commit 1ed4bef, commit ff76fc8, commit 9856ea6 (22 Sep 2021), and commit b408452 (10 Sep 2021) by Ævar Arnfjörð Bjarmason (avar
).
(Merged by Junio C Hamano -- gitster
-- in commit 62f035a, 13 Oct 2021)
Signed-off-by: Ævar Arnfjörð Bjarmason
The "help" builtin has been able to emit configuration variables since e17ca92 ("completion
: drop the hard coded list of config vars", 2018-05-26, Git v2.19.0-rc0 -- merge listed in batch #1), but it hasn't produced exactly the format the completion script wanted.
Let's do that.
We got partway there in 2675ea1 ("completion
: use 'sort -u' to deduplicate config variable names", 2019-08-13, Git v2.24.0-rc0 -- merge listed in batch #3) and d943887 ("completion
: deduplicate configuration sections", 2019-08-13, Git v2.24.0-rc0 -- merge listed in batch #3), but after both we still needed some sorting, de-duplicating and awk post-processing of the list.
We can instead simply do the relevant parsing ourselves (we were doing most of it already), and call string_list_remove_duplicates()
after already sorting the list, so the caller doesn't need to invoke "sort -u"
.
The "--config-for-completion
" output is the same as before after being passed through "sort -u
".
Then add a new "--config-sections-for-completion
" option.
Under that output we'll emit config sections like "alias
" (instead of "alias." in the --config-for-completion
output).
We need to be careful to leave the "--config-for-completion
" option compatible with users git, but are still running a shell with an older git-completion.bash
.
If we e.g. changed the option name they'd see messages about git-completion.bash being unable to find the "--config-for-completion
" option.
Such backwards compatibility isn't something we should bend over backwards for, it's only helping users who:
- Upgrade git
- Are in an old shell
- The
git-completion.bash
in that shell hasn't cached the old "--config-for-completion
" output already.
But since it's easy in this case to retain compatibility, let's do it, the older versions of git-completion.bash won't care that the input they get doesn't change after a "sort -u".
While we're at it let's make "--config-for-completion
" die if there's anything left over in "argc
", and do the same in the new "--config-sections-for-completion
" option.
With Git 2.43 (Q4 2023), git cmd -h
learned to signal which options can be negated by listing such options like --[no-]opt
".
See commit 311c8ff, commit 2a409a1, commit 652a6b1, commit e8e5d29, commit d5dc68f, commit 8dcb490, commit aa43619, commit d716512 (05 Aug 2023) by René Scharfe (rscharfe
).
(Merged by Junio C Hamano -- gitster
-- in commit 6d159f5, 25 Aug 2023)
parse-options
: show negatability of options in short help
Signed-off-by: René Scharfe
Add a "[no-]
" prefix to options without the flag PARSE_OPT_NONEG
to document the fact that you can negate them.
This looks a bit strange for options that already start with "no-", e.g. for the git option --no-name
of show-branch:
--[no-]no-name suppress naming strings
You can actually use --no-no-name
as an alias of --name
, so the short help is not wrong.
If we strip off any of the "no-"s, we lose either the ability to see if the remaining one belongs to the documented variant or to see if it can be negated.
git rev-parse
now includes in its man page:
--[no-]foo some nifty option --foo
--[no-]bar ... some cool option --bar with an argument
--[no-]baz <arg> another cool option --baz with a named argument
--[no-]qux[=<path>] qux may take a path argument but has meaning by itself