move bug report out of forum

This commit is contained in:
Joey Hess 2014-01-14 15:35:09 -04:00
parent 7569d79d92
commit 188f0bd58d
8 changed files with 0 additions and 0 deletions

View file

@ -0,0 +1,73 @@
Hello!
Can someone explain this behavior to me? System is ArchLinux with the newest prebuilt tarballs as of Jan 12th.
~ % mkdir GA1
~ % cd GA1
~/GA1 % git init
Initialized empty Git repository in /home/florian/GA1/.git/
~/GA1 (git)-[master] % git annex init
init ok
(Recording state in git...)
~/GA1 (git)-[master] % echo "GA1" > testfile
~/GA1 (git)-[master] % git annex add .
add testfile ok
(Recording state in git...)
~/GA1 (git)-[master] % git annex sync
commit ok
~/GA1 (git)-[master] % cd ..
~ % git clone GA1 GA2
Cloning into 'GA2'...
done.
~ % cd GA2
~/GA2 (git)-[master] % git annex init
init ok
(Recording state in git...)
~/GA2 (git)-[master] % git annex get .
get testfile (merging origin/git-annex into git-annex...)
(Recording state in git...)
(from origin...) ok
(Recording state in git...)
~/GA2 (git)-[master] % git annex unlock testfile
unlock testfile (copying...) ok
~/GA2 (git)-[master] % echo "GA2" > testfile
commit ok
pull origin
ok
push origin
Counting objects: 27, done.
Delta compression using up to 4 threads.
Compressing objects: 100% (16/16), done.
Writing objects: 100% (20/20), 1.88 KiB | 0 bytes/s, done.
Total 20 (delta 2), reused 0 (delta 0)
To /home/florian/GA1
f0654aa..8d0bdf9 master -> synced/master
* [new branch] git-annex -> synced/git-annex
ok
~/GA2 (git)-[master] % cd ../GA1
~/GA1 (git)-[master] % git annex sync
(merging synced/git-annex into git-annex...)
commit ok
merge synced/master
Updating f0654aa..8d0bdf9
Fast-forward
e6a/e57/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c.log | 1 +
testfile | 2 +-
2 files changed, 2 insertions(+), 1 deletion(-)
create mode 100644 e6a/e57/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c.log
ok
~/GA1 (git)-[master] % ll
total 4
drwxr-x--- 1 florian florian 6 Jan 12 13:59 e6a/
lrwxrwxrwx 1 florian florian 178 Jan 12 13:59 testfile -> .git/annex/objects/V6/ZM/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c
~/GA1 (git)-[master] % cat e6a/e57/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c.log
1389531554.951523s 1 dedf9799-0131-4773-8939-8f76294fa8e9
Where does the e6a directory comes from? I don't suspect a bug there, but in my conception of git-annex.
Thanks!
Florian

View file

@ -0,0 +1,10 @@
[[!comment format=mdwn
username="http://joeyh.name/"
ip="209.250.56.43"
subject="comment 1"
date="2014-01-13T18:57:48Z"
content="""
The log file is supposed to be stored by git-annex in the git-annex branch of the git repository. I don't know how it could end up being committed to the master branch instead.
Can you run these same commands again and reproduce the problem?
"""]]

View file

@ -0,0 +1,43 @@
[[!comment format=mdwn
username="https://www.google.com/accounts/o8/id?id=AItOawnsuhFUIfWNT-Q-C02FDaSQqceFDge5M9w"
nickname="Florian"
subject="comment 2"
date="2014-01-14T12:12:52Z"
content="""
Yes, I can always reproduce it with running \"sh test_ga.sh\" with test_ga.sh being:
set -x
mkdir GA1
cd GA1
git init
git annex init \"GA1\"
echo \"GA1\" > testfile
git annex add .
git annex sync
cd ..
git clone GA1 GA2
cd GA2
git annex init \"GA2\"
git annex get .
cat testfile
git annex unlock testfile
echo \"GA2\" > testfile
git annex sync
cd ../GA1
git annex sync
cd ..
which gives:
florian@horus ~/test % ll GA1
insgesamt 4
drwxr-x--- 1 florian florian 6 14. Jan 13:07 e6a/
lrwxrwxrwx 1 florian florian 178 14. Jan 13:07 testfile -> .git/annex/objects/V6/ZM/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c/SHA256E-s4--241dca52164d30bee7a0be8c097f02e382acff765a9d4136718a37abd437a57c
with testfile being a dead link. Complete output of the script is at http://pastebin.com/VktWAp2V
I would be glad to help furthermore.
"""]]

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="http://joeyh.name/"
ip="209.250.56.43"
subject="comment 3"
date="2014-01-14T18:55:11Z"
content="""
This script does not produce the problem for me. How did you install git-annex from the tarball?
"""]]

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="http://joeyh.name/"
ip="209.250.56.43"
subject="comment 4"
date="2014-01-14T19:00:33Z"
content="""
Actually, I was able to reproduce this using the standalone tarball and runshell.
"""]]

View file

@ -0,0 +1,10 @@
[[!comment format=mdwn
username="http://joeyh.name/"
ip="209.250.56.43"
subject="comment 5"
date="2014-01-14T19:14:35Z"
content="""
Analysis: pre-commit hook is run with `GIT_INDEX_FILE=.git/index.lock`; this seems to be causing git-annex branch commits to land in that index file instead of where they are supposed to go. Also, commits of the files in the work tree also end up adding the symlinks to the git-annex branch when this happens. Apparantly because again it's using the wrong index file when committing to the git-annex branch.
So, an easy workaround (disable pre-commit hook or make it `unset GIT_INDEX_FILE`, but I need to dig into why this env var is messing with git-annex like this.. I thought git-annex set the env var itself!
"""]]

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="http://joeyh.name/"
ip="209.250.56.43"
subject="comment 6"
date="2014-01-14T19:16:15Z"
content="""
Problem is that git-annex prepends `GIT_INDEX_FILE` to the environment; it does not overwrite it. So environment ends up with 2 copies of the var, and which will be used is undefined.
"""]]

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="http://joeyh.name/"
ip="209.250.56.43"
subject="comment 7"
date="2014-01-14T19:33:28Z"
content="""
I have fixed this in git. I will be moving this thread from the forum to the bug report tracker.
"""]]