Re: [PATCH v2 7/7] run-command: fix detaching when running auto maintenance

[Patrick Steinhardt <ps@pks.im>]

On Thu, Aug 15, 2024 at 09:13:28AM -0700, Junio C Hamano wrote:
> Patrick Steinhardt <ps@pks.im> writes:
> 
> > diff --git a/run-command.c b/run-command.c
> > index 45ba544932..94f2f3079f 100644
> > --- a/run-command.c
> > +++ b/run-command.c
> > @@ -1808,16 +1808,26 @@ void run_processes_parallel(const struct run_process_parallel_opts *opts)
> >  
> >  int prepare_auto_maintenance(int quiet, struct child_process *maint)
> >  {
> > -	int enabled;
> > +	int enabled, auto_detach;
> >  
> >  	if (!git_config_get_bool("maintenance.auto", &enabled) &&
> >  	    !enabled)
> >  		return 0;
> >  
> > +	/*
> > +	 * When `maintenance.autoDetach` isn't set, then we fall back to
> > +	 * honoring `gc.autoDetach`. This is somewhat weird, but required to
> > +	 * retain behaviour from when we used to run git-gc(1) here.
> > +	 */
> > +	if (git_config_get_bool("maintenance.autodetach", &auto_detach) &&
> > +	    git_config_get_bool("gc.autodetach", &auto_detach))
> > +		auto_detach = 1;
> 
> I think this needs somehow documented.  Something like this,
> perhaps?

Indeed, I totally forgot doing that.

> --- c/Documentation/config/maintenance.txt
> +++ w/Documentation/config/maintenance.txt
> @@ -3,6 +3,15 @@ maintenance.auto::
>  	`git maintenance run --auto` after doing their normal work. Defaults
>  	to true.
>  
> +maintenance.autoDetach::
> +	Tasks that are run via `git maintenance run --auto` by
> +	default runs in the background, if the system supports it.
> +	Setting this configuration variable to `true` explicitly
> +	asks them to run in the background, and setting it to
> +	`false` forces them to run in the foreground.  If this
> +	variable is not set, `gc.autoDetach` works as a fallback
> +	variable and behaves the same way.

This isn't entirely true. `git maintenance run --auto` will not
background, because that'd change preexisting behaviour. It also would
not make a lot of sense, because here the `--auto` trigger tells the
command to do maintenance as-needed. Coupling that with whether or not
to detach was a misdesign of git-gc(1), I think. What it does control is
whether we detach or not when automatically executing maintenance via
`prepare_auto_maintenance()`.

Anyway, my fault for not documenting it, not yours for getting it
slightly wrong.

Patrick