Merge branch 'master' of ssh://git-annex.branchable.com

This commit is contained in:
Joey Hess 2013-03-06 16:36:31 -04:00
commit 52eb62be26
5 changed files with 58 additions and 0 deletions

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="http://kaizer.se/"
nickname="ulrik.sverdrup"
subject="GPG Cipher Algo"
date="2013-03-06T04:53:06Z"
content="""
The GPG cipher algorithm is user configurable (unless annex overrides it) the best way to set it is personal-cipher-preferences. Annex' job is simply to provide enough key material for any viable GPG option (so 256 bits of random data). Also note that GPG's string-to-key \"s2k\" algorithm is an iterated application SHA-1 (also this can be configured). For a quick context, it will use a strength number (for example count=128*1024) and it will hash as many iterations as [count]/[key byte length].
"""]]

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="http://kaizer.se/"
nickname="ulrik.sverdrup"
subject="comment 6"
date="2013-03-06T05:30:16Z"
content="""
I'm not very clear, sorry. S2K is not iterative in the way for example PBKDF2 is. It will hash a very long repetition of salt + passphrase + salt + passphrase etc in one go. The repetition of this happens if more key bits than the hash digest size are needed.
"""]]

View file

@ -0,0 +1,8 @@
[[!comment format=mdwn
username="https://me.yahoo.com/a/2grhJvAC049fJnvALDXek.6MRZMTlg--#eec89"
nickname="John"
subject="comment 4"
date="2013-03-05T22:43:12Z"
content="""
Ah, thanks Joey, that explains it. An error message to that effect would be great!
"""]]

View file

@ -0,0 +1,10 @@
[[!comment format=mdwn
username="http://joeyh.name/"
nickname="joey"
subject="comment 10"
date="2013-03-06T20:35:43Z"
content="""
I've been revisiting the XMPP stuff this week, and today I fixed at least 3 problems that would keep XMPP sync from happening reliably. This would affect machines that are using XMPP and are sometimes disconnected from the net (or suspended). Nothing caused a sync to happen when restarting the assistant, or resuming from a network disconnection. This could result in both files not showing up, and file contents not being transferred, depending on the case hit.
I think it explains everything in this bug report, hopefully.
"""]]

View file

@ -0,0 +1,24 @@
I'm starting to experiment with git annex. I'd like to use it for a centralized git repo that will be checked out often, but the clones will rarely need some large binary files (used for testing). Therefore, I've set up a centralized/bare git repo and a clone of that repo using the instructions at [centralized_git_repository_tutorial](http://git-annex.branchable.com/tips/centralized_git_repository_tutorial/) and [bare_repositories](http://git-annex.branchable.com/bare_repositories/). I've added some files to the annex in the clone.
I'm struggling to understand the difference between copy, move, and get. Here's a sequence of commands:
>> git annex add shared/1bel.maegz
>> git commit -m "added first file"
>> git push
>> git annex move shared/1bel.maegz --to origin
## Now it no longer exists in my local repo
>> git annex get shared/1bel.maegz
fails.
>> git annex get shared/1bel.maegz --from origin
fails.
>> git annex copy shared/1bel.maegz --from origin
fails.
>> git annex move shared/1bel.maegz --from origin
succeeds! Now I have the file in my clone.
Each failure message is:
fatal: Could not switch to '../.git/annex/objects/W8/gZ/SHA256-s99196--62874e9b58e652c9c01e796c2bf38b2234a80e0cef95c185bb7f0857d9765df2': No such file or directory
git-annex: <file descriptor: 6>: hGetLine: end of file
How are copy, move, and get different? Which one *should* I be using to move my large data into the central (bare) repo? Will it then be available to other clones?