git/list[1] front-page[2] threads[3] people[4] search[5] about
 

Re: [PATCH] object: make add_object_array_with_mode a static function

From
Jeff King <peff@peff.net>
Date
Oct 19, 2014, 02:03 UTC
Message-ID
<20141019020319.GB17908@peff.net>
In-Reply-To
<5442F56B.8020205@ramsay1.demon.co.uk>
On Sun, Oct 19, 2014 at 12:19:07AM +0100, Ramsay Jones wrote:
Show 13 quoted lines
> I noticed that your 'jk/prune-mtime' branch also removes the only
> call to the add_object_array_with_mode() function outside of the
> object.c file; specifically commit 75ac69fa ("traverse_commit_list:
> support pending blobs/trees with paths", 15-10-2014).
> 
> This patch (which was generated using the '--histogram' option to
> format-patch), moves the function to before the definition of the
> add_object_array() function (to avoid a forward declaration), and
> makes it static.
> 
> If you need to re-roll this branch, could you please squash this
> patch into the above commit. (again, assuming you have no plans
> to add new external callers.)

That seems reasonable. Because it's a code movement, I'd actually be just as happy with it as a separate patch, where it's more obvious what is going on.

Show 5 quoted lines
> [If new external callers are very likely in the future (i.e. this
> function is an essential part of the object-array API), then it may
> well not be worth doing this. (with perhaps a note in the commit
> message? - dunno). Similar comments apply to the previous 'add_object'
> patch as well!]

I actually wondered while writing this series whether anyone actually _uses_ the mode in object_array (the new code I added sets it to the appropriate value to be on the safe side, but traverse_commit_list does not actually care about it).

Digging in the history, it looks like it is used for blob-to-blob diffs (see 01618a3, "use mode of the tree in git-diff, if <tree>:<file> syntax is used", 2007-04-22). And that still seems to be the case today. It's a shame we have to keep such complication around for one single case (especially because getting it wrong in other cases is likely to go unnoticed for years), but I think it would probably require major surgery to extract it.

I think we can take your patch a step further, though, like:
-- >8 --
Subject: [PATCH] drop add_object_array_with_mode

This is a thin compatibility wrapper around add_pending_object_with_path. But the only caller is add_object_array, which is itself just a thin compatibility wrapper. There are no external callers, so we can just remove this middle wrapper.

Noticed-by: Ramsay Jones <ramsay@ramsay1.demon.co.uk>
Signed-off-by: Jeff King <peff@peff.net>
---
I also wondered if add_pending_object_with_mode could get the same
treatment. But we _do_ use it in setup_revisions and friends when we
parse something like "HEAD:foo". It would be trivial here to call
add_pending_object_with_path instead, and actually feed "foo". That
would mean that "git rev-list --objects HEAD:foo" reported the pathname
"foo" alongside the object. Or if "foo" is a tree, all of its sub-parts
would be "foo/whatever" instead of just "whatever". That seems somewhat
sensible to me, but I would be unsurprised if it broke some weird corner
case that is expecting paths to be relative to the given tree.
 object.c | 7 +------
 object.h | 1 -
 2 files changed, 1 insertion(+), 7 deletions(-)
diff --git a/object.c b/object.c
index df86bdd..23d6c96 100644
--- a/object.c
+++ b/object.c
@@ -341,12 +341,7 @@ void add_object_array_with_path(struct object *obj, const char *name,
 
 void add_object_array(struct object *obj, const char *name, struct object_array *array)
 {
-	add_object_array_with_mode(obj, name, array, S_IFINVALID);
-}
-
-void add_object_array_with_mode(struct object *obj, const char *name, struct object_array *array, unsigned mode)
-{
-	add_object_array_with_path(obj, name, array, mode, NULL);
+	add_object_array_with_path(obj, name, array, S_IFINVALID, NULL);
 }
 
 /*
diff --git a/object.h b/object.h
index e5178a5..6416247 100644
--- a/object.h
+++ b/object.h
@@ -114,7 +114,6 @@ int object_list_contains(struct object_list *list, struct object *obj);
 
 /* Object array handling .. */
 void add_object_array(struct object *obj, const char *name, struct object_array *array);
-void add_object_array_with_mode(struct object *obj, const char *name, struct object_array *array, unsigned mode);
 void add_object_array_with_path(struct object *obj, const char *name, struct object_array *array, unsigned mode, const char *path);
 
 typedef int (*object_array_each_func_t)(struct object_array_entry *, void *);
-- 
2.1.2.596.g7379948
Previous: Ramsay JonesNext: Ramsay Jones
Message 2 of 5 in “object: make add_object_array_with_mode a static function”
  1. object: make add_object_array_with_mode a static functionRamsay Jones, Oct 18, 2014
  2. Jeff KingOct 19, 2014
  3. Ramsay JonesOct 19, 2014
  4. Junio C HamanoOct 20, 2014
  5. Jeff KingOct 20, 2014

Read the whole thread, see it on lore, or plain text.

$ cat FOOTERMessages come from the public archive at lore.kernel.org/git, fetched every hour. The front page is chosen and written each morning by an AI editor and can be wrong; the threads themselves are the record. About and API. For agents: an MCP server at https://gitlist.dev/mcp, and any thread, story or person page as Markdown by adding .md to its URL (or sending Accept: text/markdown). Details in /llms.txt.