threads / bug / 54837

Fwd: Bug in `git add -N xyz ; git stash` command

Subject: Fwd: Bug in `git add -N xyz ; git stash` command

## tl;dr

4 messages between Dec 16, 2020 and Dec 17, 2020.

replies: 3people: 2as markdown or json

Σταύρος Ντέντος· Dec 16, 2020, 10:48 UTC · lore

Hello there,

I think I have found a bug with the following series of commands:

$ git add -N xyz ; git stash error: Entry 'xyz' not uptodate. Cannot merge. Cannot save the current worktree state

It has bothered me that `git add -N xyz` "is not obvious that it is done", unless you know that you did it - and now it seems that there is some buggy behavior about that. I could've sworn that there is not even a file change on the repository (because I sync it between a VM, and after a `git pull` I saw 3 file changes, two of them being the branch head and the log branch head)

With regards, Ntentos Stavros

---------- Forwarded message ---------
From: Σταύρος Ντέντος <stdedos@gmail.com>
Date: Wed, 16 Dec 2020 at 12:32
Subject: Bug in `git add -N xyz ; git stash` command
To: git <git@vger.kernel.org>

Hello there,

I think I have found a bug with the following series of commands:

$ git add -N xyz ; git stash error: Entry 'xyz' not uptodate. Cannot merge. Cannot save the current worktree state

With regards, Ntentos Stavros

Junio C Hamano· Dec 16, 2020, 19:17 UTC · re: Σταύρος Ντέντος · lore

Re: Fwd: Bug in `git add -N xyz ; git stash` command

Σταύρος Ντέντος  <stdedos@gmail.com> writes:

Show 9 quoted lines

> I think I have found a bug with the following series of commands:
>
> $ git add -N xyz ; git stash
> error: Entry 'xyz' not uptodate. Cannot merge.
> Cannot save the current worktree state
>
> It has bothered me that `git add -N xyz` "is not obvious that it is
> done", unless you know that you did it - and now it seems that there
> is some buggy behavior about that.

Hmph, is this the same issue as what we discussed in

https://lore.kernel.org/git/CA+dzEBmh4LdN-XFS9y9YKrDMzV_7+QpDzBA2uv9Xi80PgR5NTQ@mail.gmail.com/

I wonder.

Σταύρος Ντέντος· Dec 17, 2020, 08:43 UTC · re: Junio C Hamano · lore

Re: Fwd: Bug in `git add -N xyz ; git stash` command

It seems like it, yes.

Not deciding to fix that (or not agreeing on how to fix it), should be orthogonal from the "what does the error message says" about it. I would expect something like "'xyz' is marked with intent-to-add; either stage (git add xyz) or remove the file (git ... xyz); refusing to stage."

With regards, Ntentos Stavros

On Wed, 16 Dec 2020 at 21:17, Junio C Hamano <gitster@pobox.com> wrote:

Show 18 quoted lines

>
> Σταύρος Ντέντος  <stdedos@gmail.com> writes:
>
> > I think I have found a bug with the following series of commands:
> >
> > $ git add -N xyz ; git stash
> > error: Entry 'xyz' not uptodate. Cannot merge.
> > Cannot save the current worktree state
> >
> > It has bothered me that `git add -N xyz` "is not obvious that it is
> > done", unless you know that you did it - and now it seems that there
> > is some buggy behavior about that.
>
> Hmph, is this the same issue as what we discussed in
>
> https://lore.kernel.org/git/CA+dzEBmh4LdN-XFS9y9YKrDMzV_7+QpDzBA2uv9Xi80PgR5NTQ@mail.gmail.com/
>
> I wonder.

Junio C Hamano· Dec 17, 2020, 19:42 UTC · re: Σταύρος Ντέντος · lore

Re: Fwd: Bug in `git add -N xyz ; git stash` command

Σταύρος Ντέντος  <stdedos@gmail.com> writes:

Show 5 quoted lines

> Not deciding to fix that (or not agreeing on how to fix it), should be
> orthogonal from the "what does the error message says" about it.
> I would expect something like "'xyz' is marked with intent-to-add;
> either stage (git add xyz) or remove the file (git ... xyz); refusing
> to stage."

"refusing to stash", you mean?

I think that your wording may be an improvement. Care to whip up a patch?

Thanks.