Re: [PATCH 03/15] push: reorder switch cases

Junio C Hamano wrote:

Felipe Contreras [off-list ref] writes:

quoted

We want all the cases that don't do anything with a branch first, and
then the rest.

Will help further patches.

Signed-off-by: Felipe Contreras <redacted>
---
 builtin/push.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)

diff --git a/builtin/push.c b/builtin/push.c
index f4e919450d..c19321bb9d 100644
--- a/builtin/push.c
+++ b/builtin/push.c

@@ -254,11 +254,19 @@ static void setup_default_push_refspecs(struct remote *remote)
 	int triangular = is_workflow_triangular(remote);
 
 	switch (push_default) {
-	default:
 	case PUSH_DEFAULT_MATCHING:
 		refspec_append(&rs, ":");
 		return;
 
+	case PUSH_DEFAULT_NOTHING:
+		die(_("You didn't specify any refspecs to push, and "
+		    "push.default is \"nothing\"."));
+		return;
+	default:
+	}
+
+	switch (push_default) {
+	default:

This is not quite "reorder" but split into two.

OK.

It is not yet clear
how it helps, but hopefully we'll find out why splitting the switch
into two switches is a good idea soon in a later step, so the title
needs updating to sell that aspect of the change

How about: By splitting the two kinds of modes we can return sooner in
the case the mode doesn't need to do anything else, and for the rest
that reuse quite a bit of code it can be set up after the first switch.

-- 
Felipe Contreras