Forums Bug Reports /WHO list

Hey guys

I've got a problem with updating the address list on channels, and I've noticed, mirc does NOT update the /WHO list for channels if you reconnect and have the option 'keep channels open' enabled. It does not do it even if it doesn't have the list.

First I thought it might have been because I had an older version of mirc (6.16) but today I've upgraded to see if it was fixed in there, but it still won't work.

It works fine if I close the channel windows every time before reconnecting, but that defeats the purpose of 'keep channels open' :tongue:

Thanks for your time

mIRC doesn't automatically do a /who to fill the IAL. It does
a /names and /mode .. someone plz correct me if I'm wrong.

it does do a /names upon join by default yes

i know it doesn't do a /who list on connect by default, but it does if you do

on *:join:#: { if ($nick == $me) { who $chan } }

i use it in my clonescanner to update the IAL list because /NAMES doesn't do that. Though i found out it will not update that /who list if you have 'keep channels open'

I've used that for years. I don't normally have Keep channels open
selected but I selected it and disconnected several times and that
worked every time.

If you join many channels at the same time, then that will flood you out or lag you badly..

I've never had a problem with that, but you could use a timer.

on ME:*:JOIN:#:.timerWho. $+ # 1 $chan(0) if ( # ischan) who #

As a sidenote, putting channel names directly inside a timer can cause problems and even be exploited in certain cases. The reason is that a string like #$something (which is a valid channel name) will be evaluated (try //echo -a #$me to see what I mean). If $something is $findfile() or worse, an alias that does something potentially dangerous and that the malicious channel creator knows you have, there will be a problem. The easiest way is to use $encode/$decode with the 'm' parameter:

//echo -a #$me <--

mIRCManiac <--

heh I did not know that, I do now .. thanks for that.

~ Edit ~
Maybe this should be reported as a bug?

It's actually a feature, at least for $1, $2 etc: the 5th example in the Aliases page of the help file explains.

Looking at the example in the help file, the echo above
shoud've been prefixed with a hash (#mIRCManiac) and
it wasn't (mIRCManiac).

//join #$me

mIRCManiac Illegal channel name

This one worked though
//tokenize 32 $me | echo -a #$1 <--

#mIRCManiac <--

it only works for $N tokens for others it simply evaluates them normally :S

oh i was thinking of some identifier the other day that could be used to escape whatever follows it. something like $*text evaluating to text (i chose $* since it would have to be something that can never be made a custom alias, i.e. you cant have a custom identifier named *text. $* would return $null by this logic. of course this would break existing scripts that use $*.. but maybe its about time he got rid of that? :P)

so anyway, if such an identifier was made it could be used like so: //timer 1 $chan(0) if ($* $+ # ischan) who $!v1

of course the possibility of # containing parentheses could mess around with the if condition, but maybe special behavior for this hypothetical $* could be implemented? so to clarify, $* [ $+ stuff ] = stuff (and $* is the first suitable identifier i could think of for this ). i didnt think this deserved its own post in the features forum lol

another practical example: //timer 1 3 kick # $nick

instead of the workarounds to prevent the exploits qwerty mentioned, my theoretical $* could be put to use:

//timer 1 3 kick $!* $+ # $!* $+ $nick

Personally, I think your suggested idea is not going to be that effective, because you have to escape every space-separated word that could possibly be evaluated, that way (as opposed to, for example, using a custom $escape(<whole line>) identifier which could be built with $encode+$decode), so escaping whole arbitrary lines will still be a pain.

I still think that a much more elegant solution would be a non-evaluation switch to the commands that double-evaluate by default (link)..