Some things to point out:
1. practically no clients in existance these days fail to uphold ANSI standards

2. many people these days do not know what "ANSI" even is.  for many people, "Do you want ANSI color? [Y/N]" is their first impression of the MUD world.  And we only get one first impression.

3. References to ANSI may be somewhat obsolete:  it would suffice to simply ask "Do you want color?"  If anyone has a client too ancient for ANSI, they are likely to know what you're talking about, which this comes off much friendlier to people who've never heard of it.  (Note that this applies to any MUD which prompts for ANSI anywhere, not just login)

4. Worst case scenario, sending ANSI signals to a client which doesn't support them, will make garbled gibberish appear.  Contrary to popular belief, it WON'T cause anyone's computer to explode

5. Many MUDs prompt for ANSI both on intial login, AND in character creation.  The obvious redundancy just reeks unprofessional

6. As I'm sure 30 people will immediately reply, yes, I already know you can make a trigger to auto-send the "Y" at the login prompt, as does anyone else who is reading these forums.  So please don't waste HD space reminding us for the trillionth time that triggers exist.

7. Many of the new clients have ways of identifying themselves, or at least identifying that they support certain features (mccp, pueblo, etc.) If a client falls into this category, obviously it can support ANSI; is it still necessary to prompt for ANSI from it anyway?

[Samson]

Since neither option fit what I would have answered, I filed a null vote. I opt for simply throwing a color greeting out there without prior confirmation. As you say, most if not all modern clients or telnet programs will at least know what an ANSI sequence is, even if they won't process it. Even the lowly Windows98 telnet will simply ignore them.

[Rhuarc]

Originally Posted by (erdos @ Jan. 08 2004,10:28) 4. Worst case scenario, sending ANSI signals to a client which doesn't support them, will make garbled gibberish appear.  Contrary to popular belief, it WON'T cause anyone's computer to explode

Talk about first impressions!

 Hey, why not do something original and autodetect ANSI compatibility on the client? For an example, see the mudcore engine. (www.mudcore.com)

[shadowfyr]

To put in my two cents... I remember long ago, in a reality far far away, a time people used to dial in using modems to some strange device called a BBS. It was amazing, most clients that where not pure junk would recieve a simlple 'can do ANSI?' request and reply with a 'will do ANSI' sequence of there own and no one ever had to worry about if the end user even had a clue what it bloody was. It would just end up in color.

Now, jump forward 10 years or so and we are using TCP/IP, Telnet, HTML, etc, etc. and most muds either ask the user if they want color or force them to execute a command like 'term on' to enable it. WTF? If someone wants to turn off color, then great. Otherwise why not use the existing ANSI negotiation sequence that has been around since the days of 4800baud modems to ask the client if it handles it and turn it on automatically.

The benefit is obvious, if the client works right they user will just end up seeing stuff in color. If the client's ANSI implimentation is actually broken, all you see is a few odd characters during the negotiation and the mud sends plain text after that. For the life of me, I don't get why this is so hard to do, but no one uses it. Is every Telnet application in existance actually broken and can't respond to the sequence? So much for progress if that is the case. What will be next, start making web servers that refuse to send HTML until you send a message to activate it for your specific IP?

Sorry, that was a bit of a rant, but I find it plain silly...

[Rhuarc]

That is exactly what I was referring to.

[Kylotan]

Originally Posted by (erdos @ Jan. 08 2004,10:28) 7. Many of the new clients have ways of identifying themselves, or at least identifying that they support certain features (mccp, pueblo, etc.) If a client falls into this category, obviously it can support ANSI; is it still necessary to prompt for ANSI from it anyway?

Does 'new client' mean 'pretty much every client ever made' in this context? I don't recall seeing a client that didn't send 'ANSI' as one of the terminal types if you ask for them, so you can pretty much guarantee an accurate answer that way, if you're willing to delay the greeting screen a second or two while you query the client. I'm not sure if there is a separate telnet option you can ask for that specifically deals with colour support.

[Kylotan]

Originally Posted by (shadowfyr @ Jan. 09 2004,14:50) Is every Telnet application in existance actually broken and can't respond to the sequence? So much for progress if that is the case.

You'd be surprised at some of the ways in which various versions of ZMud have been broken. My version in particular fails to follow the quite simple examples at http://www.faqs.org/rfcs/rfc1091.html and sends the terminal type list incorrectly, meaning I have to do extra server-side checks.

[Faeton]

Hoi,

I don't think a greeting like that (Do you want ansi?) is at all a greeting (as you said, it probably is confusing to most).
I prefer to have them create their character, enter the game, and then, whereever they turn out to be, place a sign with the initial hints (or have a book or whatever) which also mentions how to change color levels i.e. how to turn ansi on/off.
This doesn't ride over the newbie like a truck, and you're not pushing him into something either. Quite (user-:) friendly imho.

Regards,

-faeton

Originally Posted by (shadowfyr @ Jan. 09 2004,14:50) (Suggestion to use RFC standards to autodetect ANSI)

Amen, that's what I say too.  In fact I brought it up in these forums some months ago;  the more general idea of using the telnet standard for various things, not just ANSI negotiations.  As people who read the thread then will assure you, it was met with a combination of ignorant people saying it wouldn't work and less ignorant people pointing out that lots of MUD clients don't even support the telnet protocol (which wouldn't really matter since at worst case scenario they'd just gag them, I defy you to name one modern MUD client which doesn't at least do this much). One idiot even claimed that if you negotiated using the telnet standard, you were opening yourself to hackers :P

I think really the issue is more that the majority of "coders" wouldn't have a ghost of a clue how to even begin--- after all, this wouldn't just be yet another "copy and paste the spell_fireball function and rename it" job which seems to comprise the bulk of most coders' time!

[shadowfyr]

Originally Posted by (Kylotan @ Jan. 09 2004,15:36) Originally Posted by (shadowfyr @ Jan. 09 2004,14:50) Is every Telnet application in existance actually broken and can't respond to the sequence? So much for progress if that is the case. You'd be surprised at some of the ways in which various versions of ZMud have been broken. My version in particular fails to follow the quite simple examples at http://www.faqs.org/rfcs/rfc1091.html and sends the terminal type list incorrectly, meaning I have to do extra server-side checks.

No.. It doesn't surprise me at all that stuff in zMud is broken. Heck the specs for how to handle < and > due to MXP doesn't even follow Zugg's own specs, that state you are 'required' to replace them with &lt and &gt when MXP is used. But then it also doesn't wait for MXP negotiation before turning it on either. Wish he would either change the spec to say that the current behavior of just displaying anything that doesn't parse is *right* so MUSHclient would be changed to do that too, or Zugg fixed his. zMud has more bugs per MB than Windows 3.1 did imho. lol

Hmm. Unforunately it appears the simple sequence I was looking for is [6n. Clients that ignore text position commands will likely ignore this one as well, but it was quite commonly used in the BBS days to detect ANSI compatible terminals (any response meant it was VT-something, if not specifically ANSI), which also meant that color would probably work on it. All the complicated tests for specific terminal types is fairly new and apparently with zMud, not always believable.