On Mon, Sep 15, 2025 at 12:56 PM Patrick Steinhardt [off-list ref] wrote:

On Mon, Sep 15, 2025 at 12:17:33PM +0200, Christian Couder wrote:

quoted

On Mon, Sep 15, 2025 at 8:27 AM Patrick Steinhardt [off-list ref] wrote:

quoted

On Fri, Sep 12, 2025 at 02:40:42PM +0200, Christian Couder wrote:

quoted

quoted

quoted

+             switch (signed_commit_mode) {
+             case SIGN_ABORT:
+                     BUG("SIGN_ABORT should be handled before calling parse_one_signature()");
+                     break;

Let's be defensive and convert this into a `default:` case so that any
unhandled value will cause a BUG.

Ok, maybe something like BUG("invalid signed_commit_mode value %d",
signed_commit_mode) then?

Yeah, that should do the job.

In the V3 I just sent, I have changed the switch (...) { ... } but
there is a `default: ...` case with the BUG() instruction we
discussed. Hopefully it still does the job.

Thanks.