Re: [PATCH v2 2/3] push: Add support for pre-push hooks

[Junio C Hamano <gitster@pobox.com>]

Aaron Schrab <aaron@schrab.com> writes:

> Add support for a pre-push hook which can be used to determine if the
> set of refs to be pushed is suitable for the target repository.  The
> hook is run with two arguments specifying the name and location of the
> destination repository.
>
> Information about what is to be pushed is provided by sending lines of
> the following form to the hook's standard input:
>
>   <local ref> SP <local sha1> SP <remote ref> SP <remote sha1> LF
>
> If the hook exits with a non-zero status, the push will be aborted.
>
> This will allow the script to determine if the push is acceptable based
> on the target repository and branch(es), the commits which are to be
> pushed, and even the source branches in some cases.
>
> Signed-off-by: Aaron Schrab <aaron@schrab.com>
> ---
>  Documentation/githooks.txt |  29 ++++++++++
>  builtin/push.c             |   1 +
>  t/t5571-pre-push-hook.sh   | 129 +++++++++++++++++++++++++++++++++++++++++++++
>  transport.c                |  60 +++++++++++++++++++++
>  transport.h                |   1 +
>  5 files changed, 220 insertions(+)
>  create mode 100755 t/t5571-pre-push-hook.sh
>
> diff --git a/Documentation/githooks.txt b/Documentation/githooks.txt
> index b9003fe..d839233 100644
> --- a/Documentation/githooks.txt
> +++ b/Documentation/githooks.txt
> @@ -176,6 +176,35 @@ save and restore any form of metadata associated with the working tree
>  (eg: permissions/ownership, ACLS, etc).  See contrib/hooks/setgitperms.perl
>  for an example of how to do this.
>  
> +pre-push
> +~~~~~~~~
> +
> +This hook is called by 'git push' and can be used to prevent a push from taking
> +place.  The hook is called with two parameters which provide the name and
> +location of the destination remote, if a named remote is not being used both
> +values will be the same.
> +
> +Information about what is to be pushed is provided on the hook's standard
> +input with lines of the form:
> +
> +  <local ref> SP <local sha1> SP <remote ref> SP <remote sha1> LF
> +
> +For instance, if the command +git push origin master:foreign+ were run the

Just being curious, but why use +monospace text+ here?  Most of the
new text use `monospace text literally` instead in this patch.

> +hook would receive a line like the following:
> +
> +  refs/heads/master 67890 refs/heads/foreign 12345
> +
> +although the full, 40-character SHA1s would be supplied.

Perhaps ellipses are called for here?

    refs/heads/master 67890... refs/heads/foreign 12345...

 (the above abbreviates full 40-hexdigits for illustration purposes only)

> +If the foreign ref
> +does not yet exist the `<remote SHA1>` will be 40 `0`.  If a ref is to be
> +deleted, the `<local ref>` will be supplied as `(delete)` and the `<local
> +SHA1>` will be 40 `0`.  If the local commit was specified by something other
> +than a name which could be expanded (such as `HEAD~`, or a SHA1) it will be
> +supplied as it was originally given.
> +
> +If this hook exits with a non-zero status, 'git push' will abort without
> +pushing anything.  Information about why the push is rejected may be sent
> +to the user by writing to standard error.

s/standard error/& of the hook/; perhaps?  It is unclear who does
the writing and it can be misunderstood that git-push will write to
standard error upon seeing your hook that silently exits.

> diff --git a/builtin/push.c b/builtin/push.c
> index 8491e43..b158028 100644
> --- a/builtin/push.c
> +++ b/builtin/push.c
> @@ -407,6 +407,7 @@ int cmd_push(int argc, const char **argv, const char *prefix)
>  		OPT_BOOL(0, "progress", &progress, N_("force progress reporting")),
>  		OPT_BIT(0, "prune", &flags, N_("prune locally removed refs"),
>  			TRANSPORT_PUSH_PRUNE),
> +		OPT_BIT(0, "no-verify", &flags, N_("bypass pre-push hook"), TRANSPORT_PUSH_NO_HOOK),
>  		OPT_END()
>  	};

So to countermand this, you have to say --no-no-verify?  Wouldn't it
be more natural to introduce a --verify option that turns the bit
on, which automatically gives you --no-verify to turn it off?  A
bit in a flag word can be initialized to true before the flag word
is given to the parse_options() machinery to make the field default
to true, no?