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

Re: [PATCH 2/2] Demonstrate merge failure when a directory is replaced with a symlink.

From
Michael J Gruber <git@drmicha.warpmail.net>
Date
Jul 29, 2009, 08:29 UTC
Message-ID
<4A70086C.9070408@drmicha.warpmail.net>
In-Reply-To
<1248819198-13921-3-git-send-email-james.e.pickens@intel.com>
James Pickens venit, vidit, dixit 29.07.2009 00:13:
Show 9 quoted lines
> This test creates two directories, a/b and a/b-2, replaces a/b-2 with a
> symlink to a/b, then merges that change into another branch that
> contains unrelated changes.  Since the changes are unrelated, the merge
> should be free of conflicts, but 'git merge' gives a file/directory
> conflict.
> 
> Note that this test is almost identical to t6035, except that instead of
> replacing a/b with a symlink, it replaces a/b-2 with a symlink.  This
> test results in a merge conflict, whereas t6035 does not.

In fact they are identical: Exchange b for b-2 and vice versa everywhere and you get the same test, except for the fact that in 1/2 you "test -f" in the last step. But I'm sure that test fails at the merge step already (because of a dirty worktree), doesn't it? You should see this when running the test with -d/-v. (I'm guessing, I haven't run your test.)

Show 43 quoted lines
> 
> Signed-off-by: James Pickens <james.e.pickens@intel.com>
> ---
>  t/t6036-merge-dir-to-symlink.sh |   30 ++++++++++++++++++++++++++++++
>  1 files changed, 30 insertions(+), 0 deletions(-)
>  create mode 100755 t/t6036-merge-dir-to-symlink.sh
> 
> diff --git a/t/t6036-merge-dir-to-symlink.sh b/t/t6036-merge-dir-to-symlink.sh
> new file mode 100755
> index 0000000..020db7c
> --- /dev/null
> +++ b/t/t6036-merge-dir-to-symlink.sh
> @@ -0,0 +1,30 @@
> +#!/bin/sh
> +
> +test_description='merging when a directory was replaced with a symlink'
> +. ./test-lib.sh
> +
> +test_expect_success setup '
> +	mkdir -p a/b/c a/b-2/c &&
> +	> a/b/c/d &&
> +	> a/b-2/c/d &&
> +	> a/x &&
> +	git add -A &&
> +	git commit -m base &&
> +	rm -rf a/b-2 &&
> +	ln -s b a/b-2 &&
> +	git add -A &&
> +	git commit -m "dir to symlink"
> +'
> +
> +test_expect_failure 'checkout should not delete a/b/c/d' '
> +	git checkout -b temp HEAD^ &&
> +	test -f a/b/c/d
> +'
> +
> +test_expect_failure 'merge should not have conflicts' '
> +	echo x > a/x &&
> +	git add a/x &&
> +	git commit -m x &&
> +	git merge master'
> +
> +test_done

As in 1/2, I think the first expect_failure leaves a dirty/unexpected worktree (d missing) which causes the merge failure in the last step.

Michael
Previous: James PickensNext: Pickens, James E
Message 4 of 18 in “More symlink/directory troubles”
  1. James PickensJul 28, 2009
  2. 1/2 Demonstrate bugs when a directory is replaced with a symlink.James Pickens, Jul 28, 2009
  3. 2/2 Demonstrate merge failure when a directory is replaced with a symlink.James Pickens, Jul 28, 2009
  4. Michael J GruberJul 29, 2009
  5. Pickens, James EJul 29, 2009
  6. Michael J GruberJul 29, 2009
  7. Junio C HamanoJul 29, 2009
  8. Pickens, James EJul 29, 2009
  9. Demonstrate bugs when a directory is replaced with a symlinkPickens, James E, Jul 29, 2009
  10. Junio C HamanoJul 29, 2009
  11. Demonstrate bugs when a directory is replaced with a symlinkPickens, James E, Jul 29, 2009
  12. Linus TorvaldsJul 29, 2009
  13. Junio C HamanoJul 29, 2009
  14. Kjetil BarvikJul 29, 2009
  15. Linus TorvaldsJul 29, 2009
  16. Linus TorvaldsJul 30, 2009
  17. Junio C HamanoJul 30, 2009
  18. Junio C HamanoJul 30, 2009

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.