jps wrote:I should add to that, there is a known scenario where directory scanning will take an unbounded amount of time:
- Must be falling back to content fingerprints (i.e., stat reports all inodes as 0)
- Must have a recursive file system, where the contents are changing faster than the time it takes to scan the directory tree
In this case, because the contents of the directory are continuously changing, the content fingerprint based approach will not realise that it's scanned the directory in question before.
Yea, it looks like stat is reporting all the inodes in my home directory as 0, but ls -i also shows the correct inodes. The inodes in the .snapshot/hourly* directories are the exact same, ie inode(~/.snapshot/hourly0/sublime) == inode(~/subime) so I'd guess if those were used it would appear to sublime as just a link?
sublimator wrote:Do you work for Google ajpalkovic?
Uh, yea, I start a week or so ago.