World: r3wp

Wait a minute, I can call xclip to get the right clipboard contents...
	>> call "xclip -o"

That prints the clipboard contents into the rebol shell.

Oops, that should actually be:
	>> call "xclip -o -selection clipboard"

Doc, or anyone, do you know how to write a proxy for system/ports/input 
?

I would try to wrap a custom port wiring all methods to the console 
port then I would replace system/ports/input with that new one and 
see if the "proxy" port works. If it works, then you can intercept 
the incoming events and add your own handlers. But that's only theory, 
as much other low-level parts of REBOL, it's not documented, so I'm 
not sure it would work.

Hmm.. sounds a bit too complicated for now - might have to wait for 
another day...

but, how do you think I would replace system/ports/input ? By just 
:

	system/ports/input: my-port
?

Yes.

Ok, I'll give it a quick go to see how feasible it is. My first try 
about an hour and half ago, just using the console port looks like 
this:

console-port: open/binary/no-wait console:/
console-port/awake: func [port /local ch][
	if ch: pick port 1 [
		;print ["Console awake trapped key:" to-char ch "(" ch ")"]
		either ch = #"^V" [	; Ctrl+V

   call "xclip -o -selection clipboard" ; Three different types of clipboards: 
   "primary", "secondary" or "clipboard".
		][
			insert system/ports/output to-char ch
		]
	]
	return false ; Does not cause return from WAIT (DO-EVENTS).
	;true ; Does cause return from WAIT (DO-EVENTS).
]

insert system/ports/wait-list console-port
if error? set/any 'err try [
	do-events
][
	print mold disarm err
]
close console-port
remove find system/ports/wait-list console-port

I've tried replacing system/ports/input/awake with an awake function 
similar to above, and it doesn't seem ever to be called. If it's 
possible at all I think making a handler object with all the functions 
(open close write etc.) *might* work.

I don't really know what I'm doing with ports. I need more information. 
All I can get it to do is shutdown rebol on keypress.

I tried making a handler object with stub functions.

No joy.

awake is only called if you use WAIT (which the native console does 
not use - indeed you don't get other port events while at the console 
prompt)

i did replace system/ports/input and output and it works, i used 
that to "broadcast" what is typed at the console to other listener 
(eg. for online demos or lessons or things like that).

http://www.colellachiara.com/soft/Misc/broadcast-console.r

Is this a key logger?

lol, you can see it this way.

Gabriele, thanks, that's very interesting, I will check it out. (Aha 
 make Root-protocol, I forgot about that.)

I had a go at it again with Gabriele's way of doing it. So far I 
can only intercept input and output when enter is pressed, not at 
individual key presses. I think I can only open system/ports/input 
in lines mode, the same as the default. From memory, now, I seem 
to remember somebody having a go at this and arriving at the same 
point. (Can anybody verify that?)

Not sure I heard anything like that. The only thing I remember is 
that we got some advanced console mode, where you could use any key 
IIRC. There was even some animation done by Bo. The article was at 
rebolforces.com somewhere - not sure it could help your case though 
..

Ah, that is something different than what you need - http://www.rebolforces.com/articles/tui-dialect/
... but it reminded me to remember, how badly the windows console 
sucks :-(

system/ports/input needs to be in /lines mode IIRC... but you could 
open console:// in /binary mode, process it, and behave in lines 
mode for your custom scheme. More complicated than my example above 
for sure...

(still, ctrl-c and ctrl-v have always had a different meaning on 
terminals, so I'm not sure replacing them is a good idea. in particular, 
most likely you'd need to trap ctrl-c by trapping SIGINT... and that 
does not look like a great idea to me...)

Gabriele, could elaborate on that idea a bit? I'm having trouble 
visualizing how it would work.

(You are right about terminal signalling, but of course this is not 
the only thing this proxy port is useful for.)

in my example, you have your own port using your own scheme, and 
its sub-port which is the original system/ports/input. this sub-port 
need not be the original, you can open console:// yourself with your 
own flags (eg. binary). then you can process one char at a time, 
figure out line ending yourself, and send the entire line to the 
other side.

i think I probably need to write some code to make it more clear...

That makes it much more clear.

Hmm... more thinking and lots of questions come up.

Would your above system need WAIT ?

The only way I got console port to work is to set its awake function 
and use WAIT.

Or is there a way to set the handler in a console port?

Anyway, so it seems like you suggest to replace the default port 
at system/ports/input with an equivalent one which the system is 
expecting (ie. a direct/read/lines port), and then the subport can 
be a custom console port with binary, one char at a time stuff.

(It's unclear to me the relationship between a port and its subport. 
I know the port manages the opening/closing of its subport, but how 
can the port in lines mode transfer individual chars from its buffer 
into the subport ? Or do they both receive from the same input simultaneously 
automatically?)

when you're not in pass-thru mode, the only functions that get called 
are read and write. they both work at the binary level (read-io and 
write-io), so the conversion to lines is done by the native code 
under the hood. i don't know if you need pass-thru, though, and whether 
pass-thru will work for system/ports/input.

basically, on read you call read-io on the subport, maybe modify 
the buffer, etc. on write (for the output port), you can do the same, 
modify the buffer, then use write-io on the subport. wait on the 
subport is called by the system whenever wait on your port is called 
(that's why there's a get-sub-port function there).

Anyone know anything Apache exploits?  My site has been defaced http://www.compkarori.co.nz:8090

This was running Mindtouch deki wiki ... but now I get the same rubbish 
at any page so I suspect something is up with Apache.

you have been Hacked By Cyb3rking

do u have direct access to the server?

may need to shut off the various ports first before you can even 
get it resolved

http://tech.yahoo.com/qa/20090328150137AAzlEuc

don't understand the yahoo thing

there's only one port open to the world .. the one for web access

well, going to login to the console and have a poke around

Sorry to hear you've been attacked, Graham.

I hope it is not too much work to get it all back together....Don't 
work all night!

Looks like it was not personal -- just some automated tools seeking 
vulnerabilities.

You seem to be on a fairly recent version of Apache, but that does 
mean you may be some months behind on the security patches:
http://httpd.apache.org/security/vulnerabilities_22.html

actually, I thought Graham was on Apache 1.3*

Server id  says 2.2.3. (that may be just a reply string, not the 
real situation of course).

well, index.php has been changed

I have an idle Cheyenne running on a Linode server without a domain 
name. it's been there for a couple of months now. I was a bit surprised 
to already see bots looking for wordpress, admin pages and attempts 
to submit various scripts for injection in the access log. Well, 
I guess I shouldn't be surprised.