On Thu, Jul 31, 2025 at 03:46:05PM -0700, Junio C Hamano wrote: > Teach the unified split_string() machinery a new flag bit, > STRING_LIST_SPLIT_NONEMPTY, to cause empty split pieces omitted from s/omitted/to be &/ > the resulting string list. > > Signed-off-by: Junio C Hamano <gitster@xxxxxxxxx> > --- > string-list.c | 3 +++ > string-list.h | 3 +++ > t/unit-tests/u-string-list.c | 15 +++++++++++++++ > 3 files changed, 21 insertions(+) > > diff --git a/string-list.c b/string-list.c > index 86a309f8fb..343cf1ca90 100644 > --- a/string-list.c > +++ b/string-list.c > @@ -294,6 +294,9 @@ static int append_one(struct string_list *list, > break; > } > > + if ((flags & STRING_LIST_SPLIT_NONEMPTY) && (end <= p)) > + return 0; Okay, this is where the return value of `append_one()` starts to make sense. The condition for `end <= p` is probably overly defensive, as it shouldn't ever happen that `end < p`. We could make that a `BUG()`, but I'm not sure that's really worth it. Patrick
