Having said that: My guess would be that most of the people on 24x7 are there because they didn't close their browser on Fullpage Chat, (and thus it's fetching /msgs periodically), or using some other form of client which doesn't log them out when idle. I know I'm guilty of both.

So I'll propose some solutions: The technical one would be to 'ignore' users just fetching xml-ticker nodes in 'Other Users' and it's cache. Or, mark them extra as being 'idle', which would mean 'not actively reading posts, writing posts, sending on the CB, etc'. This would mean having the xml-tickers have their own nodetype, probably, so it could be checked upon cookie verification (wherever the 'logged in' cache gets updated, I can't seem to find it). I know not all tickers require logins, but I'm guessing that some clients at least don't bother swapping out the cookie for non-cookie tickers, so a general solution is probably better. (And wouldn't hurt any)

There's also the 'social' solution, pointing fingers at people who are on lots, at clients which are 'broken'. I don't consider this a good solution, I'd rather some technical improvements on defining whos 'there' and who isn't.

Anyway, I will fix those of my clients that are fixable (the intermud gateway), the telnet one only shows people on when they send, anyway, and the fullpage chat emulation works as the normal FPC. (Auto-refresh of /msgs)

Update: Fixed im2-intermud gateway to stop fetching /msgs after an hour of not sending anything for a particular user. (Note: no affect on the telnet interface).

C.

But hey - there's no secrets between Monks, is there? :)

The proposed solutions so far, seem are a bit extreme to me:

1. Logging in as an XML ticker makes you completely invisible, for whatever you do, be it surfing on the site or talking on the chatterbox. It is behaving less than ideal.
2. Making a special nodetype for the fullpage chat frame pages, as castaway proposes, would behave perfectly, but it seems like a lot of work to me. So I think it won't happen.

If you then add this parameter to the URL for the refresh for the pages in Fullpage Chat, then these pages fetches that are automatically done by the browser, would be ignored for other users, while manually loaded pages, would be visible. It seems perfect.

I have to admit, I rarely, if ever, look at Other Users. I wait for people to say something to me, or in the CB, before I assume they're there. (Possibly because of the MUDding background, where its also quite likely that half the 'who' list is actually gone to bed.. Especially in these days of cheap 24-hour internet connections..)

Anyfish, I think you mis-read my idea a little. The special nodetype would be for the xml-ticker nodes, the using of which could then be ignored (ie 'lasttime' wouldnt be set for users getting nodes of that type.) I dont think it would be all that difficult, either.

Fullpage chat, on the other hand, could just use the current way of being invisible, which is the logging on as an XML ticker thing, that you pointed to. *Just* for the purpose of getting the 3 frames it has /msgs, cb and other users. (Set up a second cookie for it, or something)

Specifically, logging on as an XML ticker sets a slightly different cookie. Clever clients, which hold the cookie in memory, could easily swap cookies to do one thing, and back to do another. They could just as easily not use a cookie at all, to fetch things like Other Users, CB, newest nodes etc, that don't need a login. The question is, can we persuade them to, its a social engineering thing.

If FPC gets fixed, theres still the 'other clients' ..

C.

There seem to be two issues here:

• A cosmetic issue - people look like they are here when they are not.
• A load issue - people leaving their chat client open when they are not around.

But in either case, the reality is that user education and finger pointing is only ever going to go so far in fixing the issues. There will always be people not logging out either out of laziness, forgetfulness, being distracted, etc..

Step 1 think about both issues and decide how important they really are. Step 2 if either of those issues turn out to actually be important, then find a way to solve the issues on the server side so it doesn't matter if people log out or not.

Personally, I've hacked my version of getchat.pl from the default distro so that it alarms out if I leave it alone for too long. It probably also wouldn't be too hard for somebody to do that with the framechat client also.

I had noticed that I dropped off of the "Other Users" list, even though I was obviously here, when I hadn't actively done anything monk-ish for a period of time. Now I know why.

Or just self-implement a feature in your favorite chatter client that makes you appear logged out, using the invis. monk trick. Just add a ';ticker=yes' to the end of all URI's you access via the client or do it once while doing a login op and save the cookie you get (I believe the only cookie change is a simple s/0$/1/;). This way you can still access content via your logged in user, yet the 'last here' field on your home node will not be updated (thus keeping your name off the Other Users lists).

The point is that one sees other users that are not there, not that one shows up oneself. Its a social thing..

C.

When you don't do that it appears that you are actually here when you are not

In CB one day I asked the question "so.. how does the logic for detecting monks that are in the Monestary work?" I don't remember the answer (or if I got one) but the reason I asked was a (I thought) reasonable one.

Currently the "Other Users" nodelet merely displays who is in PM without any type of metric describing their level of activity.

How much effort would there be in adding logic to keep track of the last time a user sent something to CB or submitted a node? (I'm asking this meaning is it possible and would it require a lot of effort?)

If a user became inactive there could be another nodelet titled "Sleeping Monks" or we could just remove them from the "Other Users" nodelet without any further comment until they were active again.

Part of the challenge would deciding a metric that was reasonable (i.e. not too stringent yet not too wide open) for determining when a monk was sleeping or was active. Is 3 minutes of inactivity a sleeping monk? 10 minutes? An hour?

Just food for thought....

Hey all. I did a little experiment becouse I found it interensting that I was in the "others". I left PM until my name was removed from the list. I verified it was not on the list by Not Logging in and checking the list.

Then All I ddi was log into PM and my name appeared.

So what does this mean? Well it looks like the system is logging anyone who is logged in but not in chat as an "other". It has nothing to do with the Chat.

Your experiment didn't prove or disprove anything that wasn't already pretty much understood.

"Other Users" shows all users who are currently logged in. To be logged in currently, one must interact with the www.perlmonks.org (or .com) site in some way within the past few minutes. The list decays with time; if you don't interact with the site, after a period of time, you automatically drop off the list. By "interacting" with the site, I mean browsing the site while logged in, posting to the site, searching, chatting in CB, etc. Pretty much any time your browser hits the PM site, you've interacted, and your presence on the "Other Users" list has been refreshed.

The reason that folks are talking about "Fullpage CB Chat" as a likely culprit for keeping people logged in, is that the "Fullpage CB Chat" works by periodically grabbing updates from the PM XML CB tickertape. That means as long as a user has Fullpage CB Chat running, he will always have a fresh listing on "Other Users".




Dave

I must add/comment that there are sometimes when server/net load are too high and it is practically imposible to logout and/or use any of the perlmonks options. If this is at the same time that you are going away from you computer, it is difficult to logout. So if that matters, it is another cause of some of us to seem to be here when we aren't.

Just an opinion.

You misunderstand. If you can't contact perlmonks.org to log out then then you are still effectively "not here." The only issue here is people with windows to stuff like Fullpage Chat or New Java Chat Client where they keep hitting the server and as far as PM is concerned, never actually leave.

i like much of what's already been said. I do think, an idle option, as Castaway mentioned - would be a good thing, since it's often hard to determine who's there and who isn't.

i'd queue it to say 15min's w/out having made a comment or manually clicking the refresh button or link, if possible.

but IDLE, mind you - not kicking'em out of the chatroom. The reason being, that oftentimes, people - myself included - would like to participate in the conversations, but because of the topic of the moment, may not be able to contribute. (so, i'll watch the conversations, until someone says something, which i can add to.)

i also like the opt-out feature mentioned by Zero_Flop. Not everyone wants to or has time to chat. Sometimes, one just wants to pop, grab something, and leave. (Is the Other Users category needed? I do like it, but i'm not sure, what i would use this function for - outside of maybe sending them a private, timely message.)