diff options
Diffstat (limited to 'doc/bugs/add_-J_fails_with_not_found/comment_3_9749c2722b5520e13448128b68fcfaed._comment')
| -rw-r--r-- | doc/bugs/add_-J_fails_with_not_found/comment_3_9749c2722b5520e13448128b68fcfaed._comment | 20 |
1 files changed, 20 insertions, 0 deletions
diff --git a/doc/bugs/add_-J_fails_with_not_found/comment_3_9749c2722b5520e13448128b68fcfaed._comment b/doc/bugs/add_-J_fails_with_not_found/comment_3_9749c2722b5520e13448128b68fcfaed._comment new file mode 100644 index 000000000..b2ecb4c81 --- /dev/null +++ b/doc/bugs/add_-J_fails_with_not_found/comment_3_9749c2722b5520e13448128b68fcfaed._comment @@ -0,0 +1,20 @@ +[[!comment format=mdwn + username="joey" + subject="""comment 2""" + date="2017-10-16T16:25:39Z" + content=""" +It would be good to avoid the window where the file is not present in the +working tree during ingestion, because interupting `git annex add` during +that window causes the file to go missing, with no record of it yet in git. + +I tried making the window longer by adding a 10 second sleep, and indeed +interrupting `git annex add` during the window is *bad*. + +Looking at the code, `makeLink` uses `replaceFile` to atomicallty +replace the file with the symlink. But `ingestAdd` for some reason +calls `nukeFile` before `makeLink`. I could not find any good reason +for it to do that. So, I've removed the `nukeFile`, closing the window. + +That change also fixed the "file not found" error. But I'm not sure +if it's entirely dealt with the problems raised by this bug report.. +"""]] |