Fish Protocol roundup

[Rolf Meinecke <rm codeon de>]

Hi,

recently I sent a mail to this list asking for pointer to the fish:// 
protocol. I received several mails (thanks) and had a short discussion 
with Pavel Roskin. He asked me whether I could get the discussion back 
to the list in order to make it available and archivable. So here is a 
roundup of the thread:

Pavel Machek pointed to "README.fish in vfs/."

Joerg Walter wrote:
A copy of the original documentation is in the README. I made some minor
extensions to the protocol which should be easy to figure out if you 
take a look at the perl server.


Pavel Roskin wrote:
I believe that the protocol implemented by mc allows the other side to 
be controlled by a dedicated server instead of a shell, but the fish 
server has never been implemented.

If you want to share files, then you can always use FTP.  If you want 
the traffic to be encrypted and still want mc to access it, then you 
probably could use FTP on top of IPSEC.

By the way, if you cannot easily find some software (like fish server),
you should not consider it secure even when you find it.  Security
professionals don't audit obscure software.
[..]


Then I wondered:
Hmmm,... would'nt it be possible to make a fish shell which only 
supports the commands that are required by the fish protocol and that 
can be configured like a chrooted environment (only the home dir of the 
user)? Then one could configure /bin/fish as the shell for a user and 
voila! you have a kind of fish server.

And Pavel replied:
I believe that the idea of "fish server" was something that would
interpret the shell comments, not the shell commands.  The commands
sometimes work around limitations of standard tools, such as dd.  The
comments don't have such workarounds.

Look for "command" in vfs/fish.c and you will find all you need.



Sorry for the late move but better late than never, right?

Ciao,

-rolf