comment
This commit is contained in:
parent
dcd2c95249
commit
fbd2f96b2c
1 changed files with 28 additions and 0 deletions
|
@ -0,0 +1,28 @@
|
||||||
|
[[!comment format=mdwn
|
||||||
|
username="joey"
|
||||||
|
subject="""comment 1"""
|
||||||
|
date="2021-06-14T16:45:21Z"
|
||||||
|
content="""
|
||||||
|
exporttree is already supported by both. importtree is unlikely to be
|
||||||
|
supported. It's difficult to support this without opening up a very real
|
||||||
|
possibility of data loss. If you're importing and exporting to the same
|
||||||
|
remote, what happens when there's a conflict? Eg, whatever else is writing
|
||||||
|
files to the remote writes to a file, but you locally modify the same file,
|
||||||
|
and export a tree, without importing the new version first. That
|
||||||
|
would overwrite the modified file on the remote, losing data.
|
||||||
|
|
||||||
|
All the special remotes that support exporttree+importtree are able to
|
||||||
|
avoid losing data in this scenario. (At least as well as git is able to,
|
||||||
|
there are similar races when git updates a working tree at the same time
|
||||||
|
you modify a file in the working tree, but the timing of it makes it very
|
||||||
|
unlikely, so much that noone seems to worry about it).
|
||||||
|
See also [[todo/import_tree_from_rsync_special_remote]]
|
||||||
|
|
||||||
|
I have not felt comfortable giving users a loaded gun in this case. So only
|
||||||
|
exporttree is supported, because the user doesn't expect something
|
||||||
|
else to be writing to files on that remote, or if they do, they don't
|
||||||
|
have any reason to expect git-annex to deal with it well.
|
||||||
|
|
||||||
|
Anyway, it seems to me it should be possible to install git-annex on
|
||||||
|
your pinephone and connect the repos using ssh.
|
||||||
|
"""]]
|
Loading…
Add table
Add a link
Reference in a new issue