Re: [vfs] JDBC FileSystem: planning an implementation

[Mario Ivankovits]

Wouter De Borger wrote:
but I presume this is the one file one blob version? 
 
Yes. One file one Blob. Now I have seen your Content has a number - what 
was the indention of it?
Every entry (file/directory/link) do have only one inode which is unique 
for the whole filesystem.
E.g. this makes it easy to rename/move a whole tree.
or am I wrong (the parts can be clustered with the link field, but
that would require a lot of duplicate inode data?)
 
The "link" field was meant to allow inserting the same inode at another 
place in the structure

e.g.
/root/dir1/file1.txt
/root/dir2/file2.txt
/root/dir2/file3.txt -> /root/dir1/file1.txt
which means file3.txt is the same as file1.txt e.g. its link_inode 
points to the same inode as file1.txt.

This is called "hard link" and only possible within the same filesystem 
- in other word its topmost parent_inode has to be the same as the link 
destination.

But hey. For sure we wont need this for version 1.0 - Maybe I carried 
too far ;-)

---
Mario
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]