I went ahead and integrated plaintext mssp into my crawler using the MSSP-REQUEST command, with the following data format:

\r\nMSSP-REPLY-START\r\nvariable\tvalue\r\nvariable\tvalue\r\nMSSP-REPLY-END\r\n

Any muds available for debugging?

We've seen already what happens when somebody unilaterally rushes forward like this – can we maybe avoid doing it again?

godwars2.org 3001

It's mostly filled with static dummy values, but there should be enough to test.

The telnet version is a pretty big success all things considered, and an actual implementation is often the best way to work out the kinks.


Can't connect.

Well, so long as you don't come back later telling us it's annoying to change things because you're already spent time implementing, and so long as you don't consider it set in stone because you wrote some code, I'm fine with tests…

You can't connect, or your crawler can't?

Connection works fine for me, typing "MSSP-REQUEST" then displays the appropriate data.

Hmmm let's see…





Should the command be "MSSP-PLAINTEXT" or "MSSP-REQUEST"? Do I still need the text "MSSP PLAINTEXT" somewhere on the login screen?

Connection is refused, port 3000 works, so I assume it's a firewall issue and you connect from localhost ?

The command should be MSSP-REQUEST.

The "MSSP PLAINTEXT" trigger is still under discussion, the consensus so far seems to be that Muds should register as supporting plaintext mssp so muds don't have to uglify their login screen.

It doesn't strike me as too bad to print it and then immediately issue a clear-screen request; the players wouldn't see it…

They would on gmud and various other mud clients.

There are clients that don't support ANSI/VT100 screen clearing? Good to know, I guess.

yeah, they're called mud clients :p

All MUD clients I've ever used can clear the screen even if they don't support full cursor control. I guess that means I haven't used gmud :tongue:

\e\e[2J will mess up tt and tf split mode, \e[1J should be alright though.

I'd suggest using black text:

\e[0;30m\r\nMSSP PLAINTEXT\r\n\e[0m

That should work for over 90% of the clients out there.

Sending black text is wrong for at least two reasons:

- you don't know that the client supports ANSI color codes, or any color codes for that matter
- you don't know that the background is actually black

While I understand the intent behind having a banner-sent "i do mssp", this
assumes that crawler writers are going to be polite.

I think that assuming crawler writers will be polite is excessively optimistic.

Realistically, some crawlers are going to be coded that bang away at every
mud they know about. Is it something to encourage? Dunno, maybe not.
But I think maybe we should keep things in perspective.

We're trying to avoid crawlers clogging up log files, right? Let's work out
how likely this is.

* Muds I'm used to tend to log failed and successful logins, not just a connection,
a username, and a disconnection. Am I sheltered? Is the logging norm to
catch username input even if no password is issued?

* Muds I'm used to get quite a lot of nonsense logins anyway, from curious
passersby, testers, what have you. How much more random log noise will
crawlers generate, assuming muds frequently do this "i need to know who
even touches my port" logging?

* How likely is it that if a crawler checks in more than once a day with a mud,
and this is logged at the "i need to know who enters anything at the username"
level, and someone is watching this, that they will actually be unable to
cope with it?

I think there's ways to send out this "i do mssp" stuff, it's kinda neat what people
are coming up with. But the premise here, apparently, that crawlers will be polite in the
first place, and that log-everything types can't handle frequent pokes (assuming
the case of frequent pokes from crawlers materialize).

What the community comes up with is vox dei, but honestly, I think there's better
places to spend cycles.

-Crat

This whole "I do MSSP" triggering and plastering text on the login screen becomes entirely moot if the crawlers are guided by the option chosen when an admin registers their initial listing. I envision 3 options:

1. The MUD supports plaintext MSSP logins. So the crawler should reach the login prompt and issue MSSP-REQUEST and things proceed from there.
2. The MUD supports telopt MSSP. The crawler should send the telnet codes *AND* should expect to get back a response that uses opcodes.
3. The MUD supports neither and the admin fills in data as usual. The crawlers leave these places alone.

I think option 3 has been largely ignored by sites (including our own) that are considering this. It could take months or years before enough games have adopted this to assume it's the only way to get the information.

I think Samson's suggestion is probably the best route to handle things, for both protocol versions. It would also alleviate crawlers un-necessarily requesting MSSP data from every MUD while doing connection checks, and let them only request it where they know it is provided.

A MUD supporting telopt MSSP automatically sends IAC WILL MSSP to every connecting client, same as with MCCP, so there should only be need for a plaintext mssp checkbox unless the crawler is simply going to send MSSP-REQUEST.

Given the '-' makes it an invalid username on most muds I doubt it'll spam many log files.

That's fine. I was simply trying to illustrate the options if they're made available.

The hyphen would generally make the username invalid on most Dikus. But that doesn't mean it won't spam log files either. For instance, Smaug logs all inbound connections even if they don't offer a username. So they'll even see the inbound connection for one that's using the telopt method. AFKMud won't but only because I blocked that until after the connection has at least offered up some kind of username since there used to be a LOT of people connecting to the login screen but never going past that on Alsherok. We got tired of seeing the log messages :)

Alright, updated the protocol page: