Am 14.10.22 um 17:40 schrieb Ævar Arnfjörð Bjarmason:
> Add a run_command_sv_opt() convenience wrapper for
> run_command_v_opt(), as noted in the API documentation this is for the
> common case of wanting to construct a "struct strvec" to pass to
> run_command_v_opt(), and as it's a one-shot to strvec_clear() it
> afterwards.
Interesting idea. It wastes memory by allocating the argument vector
twice, but for most call-sites this won't be noticeable. I suspect it
might fit at least some of the use cases for run_command_l_opt() better,
e.g. those that previously used a strvec or that need some flexibility.
> diff --git a/run-command.h b/run-command.h
> index 2b1fe3cde5c..639cee4f4fb 100644
> --- a/run-command.h
> +++ b/run-command.h
> @@ -151,7 +151,7 @@ struct child_process {
>
> /**
> * The functions: child_process_init, start_command, finish_command,
> - * run_command, run_command_l_opt, run_command_v_opt,
> + * run_command, run_command_l_opt, run_command_v_opt, run_command_sv_opt,
> * child_process_clear do the following:
> *
> * - If a system call failed, errno is set and -1 is returned. A diagnostic
> @@ -262,6 +262,24 @@ int run_command_v_opt(const char **argv, int opt);
> LAST_ARG_MUST_BE_NULL
> int run_command_l_opt(int opt, ...);
>
> +/**
> + * The run_command_sv_opt() function is a wrapper for
> + * run_command_v_opt(). It takes a "struct strvec *args" which
> + * similarly to run_command() (but not run_command_sv_opt()) will be
^^^^^^^^^^^^^^^^^^^^
Do you mean run_command_v_opt()?
René
