On Mon, May 19, 2025 at 03:51:19AM -0400, Jeff King wrote: > On Sun, May 18, 2025 at 11:57:07PM +0800, shejialuo wrote: > > > In "add_entry", we accept "insert_at" parameter which must be either -1 > > (auto) or between 0 and `list->nr` inclusive. Any other value is > > invalid. When caller specify any invalid "insert_at" value, we won't > > check the range and move the element, which would definitely cause the > > trouble. > > > > However, we only use "add_entry" in "string_list_insert" function and we > > always pass the "-1" for "insert_at" parameter. So, we never use this > > parameter to insert element in a user specified position. Let's delete > > this parameter. If there is any requirement later, we need to use a > > better way to do this. > > We can see from looking at the code that removing this will not change > the behavior. But that always makes me wonder why it was there in the > first place, and whether we might ever want it. > Yes, I agree. Actually, in my first implementation, I didn't realise that this is redundant. However, when inspecting the code carefully, I find out this is useless. > The answer in this case is that we used to have another function, > string_list_insert_at_index(), which used the extra insert_at parameter. > The idea being that you could call string_list_find_insert_index(), > decide whether there was something already there, and then insert > without repeating the binary search. > > But you can see in callers like 63226218ba (mailmap: use higher level > string list functions, 2014-11-24) that this was not really that useful > (in that commit we just try to insert and check the util pointer to see > if we need to add the auxiliary structure). > > So the function went away in f8c4ab611a (string_list: remove > string_list_insert_at_index() from its API, 2014-11-24), and I suspect > we won't need it again. (Also, I think these days we'd probably use a > strmap instead anyway). > Thanks for the hint. By seeing this commit, I totally understand the history. Because we delete `string_list_insert_at_index`, we simply call "add_entry" by specifying "auto" mode and somehow we don't delete the legacy check in "add_entry". But I have one question: should I include the information in the commit message? I feel doing this would be chaty. But I somehow think we should do this. Thanks, Jialuo
