Added a comment
This commit is contained in:
parent
ee9b2728e8
commit
a6ee9f4354
1 changed files with 16 additions and 0 deletions
|
@ -0,0 +1,16 @@
|
|||
[[!comment format=mdwn
|
||||
username="http://joeyh.name/"
|
||||
ip="209.250.56.87"
|
||||
subject="comment 1"
|
||||
date="2013-12-15T19:38:48Z"
|
||||
content="""
|
||||
If you don't trust a remote repository, then you should either
|
||||
|
||||
a) Not use that repository at all, because its malicious owner could put any evil file he wants in it with an entirely correct hash.
|
||||
|
||||
b) Make it a gcrypt remote so all content stored on it is encrypted. Decrypting it will include validating that you get out what you originally put in.
|
||||
|
||||
So these scenarios are not good arguments for validating every file after it's downloaded.
|
||||
|
||||
If it were possible to do a rolling checksum as part of the download, rather than needing to pull the entire file back off disk and checksum it, I'd do so. But it's generally not; for example when git-annex is downloading a file using rsync it may resume part way through a previous interrupted download, and rsync is storing the file to disk, not streaming it to git-annex.
|
||||
"""]]
|
Loading…
Reference in a new issue