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

Re: 2.37.2 can't "git pull" but 2.18.0 can

From
Johannes Schindelin <johannes.schindelin@gmx.de>
Date
Sep 7, 2022, 12:59 UTC
Message-ID
<s46p34qn-rq84-20q6-nr36-594sos6q5qq9@tzk.qr>
In-Reply-To
<CA+4x=b_KP4rMH-EDmVCD_4LHPiLx32J2OJC6Mm5xpQ-QUcSKuA@mail.gmail.com>
Hi Lana,
as per usual, replies inline.
On Tue, 6 Sep 2022, Lana Deere wrote:
Show 48 quoted lines
> The 'git ls-remote origin' command on both produces about 3600 lines
> of output which appears to consist of all the branches in our repo.
> The two outputs are identical according to diff.  Both include a
> single mention of the switch-to-qt5 branch,
>
> $ diff /tmp/ls-remote*
> [no output here]
> $ grep switch-to-qt5 /tmp/ls-remote.*
> /tmp/ls-remote.2.18:6a9363081d05c313ba6a6ac59183193f1340bb1f
> refs/heads/feature/switch-to-qt5
> /tmp/ls-remote.2.37:6a9363081d05c313ba6a6ac59183193f1340bb1f
> refs/heads/feature/switch-to-qt5
>
> The 'git config -l' from both versions is almost the same - two lines
> have moved around.
> $ diff /tmp/config*
> 1,2d0
> < filter.lfs.required=true
> < filter.lfs.clean=git-lfs clean -- %f
> 4a3,4
> > filter.lfs.required=true
> > filter.lfs.clean=git-lfs clean -- %f
>
> $ cat /tmp/config.2.37
> filter.lfs.smudge=git-lfs smudge -- %f
> filter.lfs.process=git-lfs filter-process
> filter.lfs.required=true
> filter.lfs.clean=git-lfs clean -- %f
> user.name=Lana Deere
> user.email=lana.deere@gmail.com
> filter.lfs.clean=git-lfs clean -- %f
> filter.lfs.smudge=git-lfs smudge -- %f
> filter.lfs.process=git-lfs filter-process
> filter.lfs.required=true
> color.branch=false
> color.diff=false
> color.grep=false
> color.interactive=false
> color.pager=false
> color.showbranch=false
> color.status=false
> color.ui=false
> core.repositoryformatversion=0
> core.filemode=true
> core.bare=false
> core.logallrefupdates=true
> remote.origin.url=http://lana@githost:7990/scm/dp/sw.git
> remote.origin.fetch=+refs/heads/master:refs/remotes/origin/master

At first I thought that this would be the root cause: `feature/switch-to-qt5` is not included in the refs to fetch.

But then I added a test case for that specific scenario:
-- snip --
diff --git a/t/t5520-pull.sh b/t/t5520-pull.sh
index 081808009b2..6e6ddeb7e63 100755
--- a/t/t5520-pull.sh
+++ b/t/t5520-pull.sh
@@ -218,6 +218,17 @@ test_expect_success 'fail if upstream branch does not exist' '
 	test_cmp expect file
 '

+test_expect_success 'fetch upstream branch even if refspec excludes it' '
+	git branch tirili &&
+	git branch tirili2 &&
+	git init -b tirili downstream &&
+	git -C downstream remote add -t tirili origin "file://$(pwd)/.git" &&
+	git -C downstream config branch.tirili.remote origin &&
+	git -C downstream config branch.tirili.merge refs/heads/tirili2 &&
+	git -C downstream pull 2>err &&
+	! grep "configuration specifies to merge" err
+'
+
 test_expect_success 'fail if the index has unresolved entries' '
 	git checkout -b third second^ &&
 	test_when_finished "git checkout -f copy && git branch -D third" &&

-- snap --

And that test case passes!

The reason is that we specifically add the ref that needs to be merged to
the list of refs to be fetched:
https://github.com/git/git/blob/v2.37.2/builtin/fetch.c#L605-L614

Now, clearly it is not quite working as intended in your scenario. The
message you pasted is produced by the code in
https://github.com/git/git/blob/v2.37.2/builtin/pull.c#L421-L494, which is
only entered if there are no entries in `.git/FETCH_HEAD` except
`not-for-merge` ones.

Lana, would you mind pasting the contents of `.git/FETCH_HEAD` just after
a failed `git pull`?

Ciao,
Johannes

> branch.master.remote=origin
> branch.master.merge=refs/heads/master
> branch.feature/switch-to-qt5.remote=origin
> branch.feature/switch-to-qt5.merge=refs/heads/feature/switch-to-qt5
>
>
> .. Lana (lana.deere@gmail.com)
>
> On Fri, Sep 2, 2022 at 4:16 PM brian m. carlson
> <sandals@crustytoothpaste.net> wrote:
> >
> > On 2022-09-02 at 19:27:55, Lana Deere wrote:
> > > I'm testing an upgrade to git 2.37.2 from the current version we're
> > > using of 2.18.0.  When I try to pull in my development tree, 2.37.2
> > > gives me an error but 2.18.0 things all is fine:
> > >
> > > $ /tools/linux-x86_64/git/2.37.2/bin/git pull
> > > Your configuration specifies to merge with the ref
> > > 'refs/heads/feature/switch-to-qt5'
> > > from the remote, but no such ref was fetched.
> > >
> > > $ /tools/linux-x86_64/git/2.18.0/bin/git pull
> > > From http://githost:7990/scm/dp/sw
> > >  * branch                  feature/switch-to-qt5 -> FETCH_HEAD
> > > Already up to date.
> > >
> > > Anyone have any ideas about this?  All I could find on google was a
> > > suggestion that the "no such ref" message indicates the remote branch
> > > was deleted, but that's not the case here.
> >
> > Can you provide the output of `git ls-remote origin` (assuming that's
> > the remote you're using) and `git config -l` (the latter with both
> > versions)?  I don't know of any reason why Git 2.37 should be broken in
> > this regard, but I suspect that there's a difference in configuration
> > between the two leading to this.
> > --
> > brian m. carlson (he/him or they/them)
> > Toronto, Ontario, CA
>
Previous: Lana DeereNext: Lana Deere
Message 4 of 31 in “2.37.2 can't "git pull" but 2.18.0 can”
  1. Lana DeereSep 2, 2022
  2. brian m. carlsonSep 2, 2022
  3. Lana DeereSep 6, 2022
  4. Johannes SchindelinSep 7, 2022
  5. Lana DeereSep 7, 2022
  6. Jeff KingSep 8, 2022
  7. Jeff KingSep 3, 2022
  8. Lana DeereSep 6, 2022
  9. Đoàn Trần Công DanhSep 7, 2022
  10. Lana DeereSep 7, 2022
  11. Jeff KingSep 7, 2022
  12. Lana DeereSep 7, 2022
  13. Jeff KingSep 7, 2022
  14. Lana DeereSep 8, 2022
  15. Jeff KingSep 8, 2022
  16. 0/2 v2 protocol can't "git pull" with restricted refspecJeff King, Sep 8, 2022
  17. 1/2 fetch: stop checking for NULL transport->remote in do_fetch()Jeff King, Sep 8, 2022
  18. 2/2 fetch: add branch.*.merge to default ref-prefix extensionJeff King, Sep 8, 2022
  19. Junio C HamanoSep 8, 2022
  20. Junio C HamanoSep 8, 2022
  21. Jeff KingSep 9, 2022
  22. Junio C HamanoSep 9, 2022
  23. Jeff KingSep 11, 2022
  24. Lana DeereSep 9, 2022
  25. Junio C HamanoSep 9, 2022
  26. Lana DeereSep 12, 2022
  27. Jeff KingSep 13, 2022
  28. Johannes SchindelinSep 5, 2022
  29. Lana DeereSep 6, 2022
  30. Johannes SchindelinSep 7, 2022
  31. Lana DeereSep 7, 2022

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.