Re: [PATCH 1/3] builtin-status: submodule summary support

[Johannes Sixt <j.sixt@viscovery.net>]

Ping Yin schrieb:
> +static void wt_status_print_submodule_summary(struct wt_status *s)
> +{
> +	struct child_process sm_summary;
> +	const char *argv[] = {
> +		"submodule",
> +		"summary",
> +		"--cached",
> +		"--for-status",
> +		"--summary-limit",
> +		summary_limit,

Where is summary_limit? Did you split this patch series incorrectly?

> +		s->amend ? "HEAD^" : "HEAD",

Any chance that we avoid "HEAD^" here? Perhaps we have the SHA1 around
somewhere? I fear that our MSYS bash mangles it into "HEAD". I shall test it.

BTW, you don't mention the prerequisites of this series. I assume it
builds on top of your "[PATCH v5 0/5] git-submodule summary" series.

> +	sm_summary.no_stderr = 1;

Why this? If the submodule summary has errors we certainly want to see them.

> @@ -321,6 +349,9 @@ void wt_status_print(struct wt_status *s)
>  	}
>  
>  	wt_status_print_changed(s);
> +	// must flush s->fp since following call will write to s->fp in a child process
> +	fflush(s->fp);
> +	wt_status_print_submodule_summary(s);

Hmm. Aren't you unconditionally spawning "git submodule summary" for each
git-status/git-commit?

-- Hannes