Junio C Hamano [off-list ref] writes:

Nicely done.

quoted

+		if (!capture_command(&cp, &sb, 0) && sb.len) {

...
So, while it is not wrong per-se, I do not think we need to check
revname[0] here.  The helper never returns a non-NULL pointer that
points at an empty string, right?

On the other hand, if we dropped the "&& sb.len" check in the helper
function to be more faithful to the original, then we must check
revname[0] for an empty string.

Ah, ignore all of the above.  This will all be discarded in the next
step [4/4], as far as I can tell.  Perhaps we should drop this step
and get directly to it, making the result a three-patch series
instead, then, no?