followup
This commit is contained in:
parent
a8f2f29da1
commit
0fa04f4470
2 changed files with 19 additions and 0 deletions
|
@ -1,3 +1,5 @@
|
|||
I might be wrong but given the observation that adjust --unlock took 26.04s user 33.95s system 31% cpu 3:12.36 total to finish operation on an annex (on smaug, btrfs filesystem) with a relatively small (~400) number of relatively large files, I guess no reflink copying was done.
|
||||
|
||||
[[!meta author="yoh"]]
|
||||
|
||||
> [[dup|done]] --[[Joey]]
|
||||
|
|
|
@ -0,0 +1,17 @@
|
|||
[[!comment format=mdwn
|
||||
username="joey"
|
||||
subject="""comment 1"""
|
||||
date="2016-06-09T19:44:14Z"
|
||||
content="""
|
||||
Git's smudge interface requires git-annex to output the content to stdout;
|
||||
git writes it to the file. So, git-annex does not have an opportunity to
|
||||
efficiently copy the file.
|
||||
|
||||
(All file copies in git-annex use cp --reflink)
|
||||
|
||||
My proposed extended smudge interface at
|
||||
<http://thread.gmane.org/gmane.comp.version-control.git/294425>
|
||||
lets the smudge filter write the file to disk, and then git-annex could
|
||||
use more efficient methods. Since this is already discussed in
|
||||
[[todo/smudge]] I think I'll close this bug report as a duplicate.
|
||||
"""]]
|
Loading…
Reference in a new issue