On Mon, Dec 25, 2017 at 07:26:27PM +0100, Igor Djordjevic wrote:
> But I`ve noticed that "--porcelain=v2" output might still be buggy -
> this is what having both files staged shows:
>
> $ git status --porcelain=v2
> 2 R. N... 100644 100644 100644 12f00e90b6ef79117ce6e650416b8cf517099b78 12f00e90b6ef79117ce6e650416b8cf517099b78 R100 new-file original-file
>
> ..., where having old/deleted file unstaged, and new/created file
> staged with `git add -N` shows this:
>
> $ git status --porcelain=v2
> 1 .R N... 100644 100644 100644 12f00e90b6ef79117ce6e650416b8cf517099b78 12f00e90b6ef79117ce6e650416b8cf517099b78 new-file
>
> So even though unstaged value is correctly recognized as "R" (renamed),
> first number is "1" (instead of "2" to signal rename/copy), and both
> rename score and original file name are missing.
>
> Not sure if this is a bug, but it seems so, as `git status` "Porcelain
> Format Version 2"[1] says the last path is "pathname in the commit at
> HEAD" (in case of copy/rename), which is missing here.
Yeah v2 looks problematic. The way the document is written, it's not
prepared to deal with a rename pair coming from comparing the index
(with intent-to-add entries) with worktree, only from comparing with
HEAD. So either we could ajust v2 semantics slightly like this
diff --git a/Documentation/git-status.txt b/Documentation/git-status.txt
index 81cab9aefb..3da10020aa 100644
--- a/Documentation/git-status.txt
+++ b/Documentation/git-status.txt
@@ -309,13 +309,13 @@ Renamed or copied entries have the following format:
of similarity between the source and target of the
move or copy). For example "R100" or "C75".
<path> The pathname. In a renamed/copied entry, this
- is the path in the index and in the working tree.
+ is the path in the index.
<sep> When the `-z` option is used, the 2 pathnames are separated
with a NUL (ASCII 0x00) byte; otherwise, a tab (ASCII 0x09)
byte separates them.
- <origPath> The pathname in the commit at HEAD. This is only
- present in a renamed/copied entry, and tells
- where the renamed/copied contents came from.
+ <origPath> The pathname in the commit at HEAD or in the worktree.
+ This is only present in a renamed/copied entry, and
+ tells where the renamed/copied contents came from.
--------------------------------------------------------
Unmerged entries have the following format; the first character is
The problem is, you cannot know if it's a rename from HEAD or from
worktree with this updated v2 (or perhaps you could because HEAD name
should be all zero?).
Or we disable rename-from-worktree when porcelain v2 is requested (and
optionally introduce v3 to support it). Jeff, any preference?
--
Duy
