[PATCH] string_list_append: always set util pointer to NULL

Jeff King· Feb 12, 2011, 05:18 UTC · lore

It is not immediately obvious that the util field may contain random bytes after appending an item. Especially since the string_list_insert* functions _do_ explicitly zero the util pointer.

This does not appear to be a bug in any current git code, as all callers either fill in the util field immediately or never use it. However, it is worth it to be less surprising to new users of the string-list API who may expect it to be intialized to NULL.

Signed-off-by: Jeff King <peff@peff.net>
---
This bit me in some new code I was writing. I consider it a bug fix, but
even if you don't, I think it is still an API improvement. :)

 string-list.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

Show changes to string-list.c +1 −0

diff --git a/string-list.c b/string-list.c
index 9b023a2..5168118 100644
--- a/string-list.c
+++ b/string-list.c
@@ -153,6 +153,7 @@ struct string_list_item *string_list_append(struct string_list *list, const char
 	ALLOC_GROW(list->items, list->nr + 1, list->alloc);
 	list->items[list->nr].string =
 		list->strdup_strings ? xstrdup(string) : (char *)string;
+	list->items[list->nr].util = NULL;
 	return list->items + list->nr++;
 }

-- 
1.7.4

Junio C Hamano· Feb 12, 2011, 07:43 UTC · re: Jeff King · lore

Re: [PATCH] string_list_append: always set util pointer to NULL

Jeff King <peff@peff.net> writes:

Show 14 quoted lines

> It is not immediately obvious that the util field may
> contain random bytes after appending an item. Especially
> since the string_list_insert* functions _do_ explicitly zero
> the util pointer.
>
> This does not appear to be a bug in any current git code, as
> all callers either fill in the util field immediately or
> never use it. However, it is worth it to be less surprising
> to new users of the string-list API who may expect it to be
> intialized to NULL.
>
> Signed-off-by: Jeff King <peff@peff.net>
> ---
> This bit me in some new code I was writing. I consider it a bug fix, but ...

Of course; the "Especially since" part is very convincing.

patchstring_list_append: always set util pointer to NULL