On Fri, Mar 07, 2025 at 01:32:49PM -0800, Elijah Newren wrote: > Makes sense...but doesn't the second patch also fix this issue without > the first patch being needed? It does, but the mechanism is pretty round-about. (From a quick glance we'll turn the empty pattern "" into "/" which won't match anything, and thus won't contribute to the jump list). But there are a couple of reasons to keep this patch. Most importantly, it hardens us against potential future regressions here with the empty pattern. And it makes dealing with that case much more explicit by throwing those patterns out before they make their way to the backends instead of the quirk above. Thanks, Taylor
