Re: [PATCH v2 1/4] run-command: add new check_command helper

[Johannes Sixt <j6t@kdbg.org>]

Am 02.04.2013 12:31, schrieb Felipe Contreras:
> And persistent_waitpid() to recover the information from the last run.

I'm not a fan of this new API, because it looks like a workaround
for a problem that should have been solved in a cleaner way. But if
we can't avoid it, please also add a paragraph to
Documentation/technical/api-run-command.txt

> +int check_command(struct child_process *cmd)
> +{
> +	int status;
> +	pid_t waiting;
> +
> +	if (cmd->last_status.valid)
> +		return 0;
> +
> +	while ((waiting = waitpid(cmd->pid, &status, WNOHANG)) < 0 && errno == EINTR)
> +		; /* nothing */
> +
> +	if (!waiting)
> +		return 1;
> +
> +	if (waiting == cmd->pid) {
> +		cmd->last_status.valid = 1;
> +		cmd->last_status.status = status;
> +		return 0;
> +	}
> +
> +	if (waiting > 0)
> +		die("BUG: waitpid reported a random pid?");
> +
> +	return 0;
> +}
> +
>  static void prepare_run_command_v_opt(struct child_process *cmd,
>  				      const char **argv,
>  				      int opt)
> @@ -729,7 +770,7 @@ error:
>  int finish_async(struct async *async)
>  {
>  #ifdef NO_PTHREADS
> -	return wait_or_whine(async->pid, "child process");
> +	return wait_or_whine(cmd, async->pid, "child process");

This breaks the NO_PTHREADS build because cmd is undeclared. Perhaps
this on top:

diff --git a/run-command.c b/run-command.c
index a9fa779..a02ef62 100644
--- a/run-command.c
+++ b/run-command.c
@@ -230,7 +230,7 @@ static pid_t persistent_waitpid(struct child_process *cmd, pid_t pid, int *statu
 {
 	pid_t waiting;
 
-	if (cmd->last_status.valid) {
+	if (cmd && cmd->last_status.valid) {
 		*status = cmd->last_status.status;
 		return pid;
 	}
@@ -771,7 +771,7 @@ int start_async(struct async *async)
 int finish_async(struct async *async)
 {
 #ifdef NO_PTHREADS
-	return wait_or_whine(cmd, async->pid, "child process");
+	return wait_or_whine(NULL, async->pid, "child process");
 #else
 	void *ret = (void *)(intptr_t)(-1);