Hi jim,
 jb> Date:   1997-01-31 22:29:19
 jb> From:   jim barchuk
 jb> To:     Trev Roydhouse
 jb> Subj:   An Opus rising ...
 TR> Tonight I received an email response from Wynn Wagner III (the copyright
 TR> holder) giving his permission for mounting such an effort. He also
 TR> wondered whether anyone was still interested in Opus ... a valid enough
 TR> query.
 jb> ...interface. I don't know if you've looked into this,
 jb> Wynn, I've heard that KA9Q can do it but haven't tried, but
 jb> if you could induce Opus to respond to a PPP connection and
 jb> serve HTML, I can virtually guarantee folks'd be piling
 jb> back in droves.
 jb> _ _ _ O / _ _ _ _ _ _ _ _ _ _ _ _ _
Daresay they would; might even start getting some callers back :)  But it's 
such a massive job - not just PPP, but then integrating an IP stack to run 
servers for say UDP ping, maybe finger, then TCP servers like FTP and HTTP.
Even the memory requirements in 640K DOS space are daunting, without any room 
much left for Opus itself .. I found one (gopher+ftp) server version of KA9Q 
that _required_ use of a CGA card, with RAM mapped in up to b7ff (736K)!
As for KA9Q .. a truly marvellous monster.  But it's vast, even compiled for 
FTN use (leaving out all the X.25 and ham radio protocols and interfaces etc 
that it was originally developed for).  I'm not sure whether Mark Blakely has 
published source of his versions or not, but looking at the last publicly 
released sources I saw in '95, even incorporating just PPP, IP, TCP sockets, 
and its multithreading kernel (needed to service multiple sockets) would run 
to at least 350K with buffers, _plus_ whichever servers you want to run.
 jb> -Nothing- would please me more than to see Opus do PPP. The
Trouble is, just doing PPP doesn't get you that far - it only provides data 
link, transport and some network services for higher layer servers, like FTP, 
Telnet and (presumably, as you're talking of use with browsers) HTTP.
 jb> point of my post was to describe that there really is no
 jb> parallel between OEC and HTML, that once the PPP connection
 jb> is made  the interpretation is done by the -browser- rather
 jb> than the server.
Yes, but only if you're connected to an HTTP server dishing out HTML, images 
and script transactions etc as requested, and handling connections with other 
URLs from selected links, even assuming all local (file://) references.
So do you add an HTTP server to Opus (along with ALL the lower layer stuff) 
or do you encourage Opus, on first logging in, to hand off to a process that 
can load PPP and a TCP/IP stack, then servers for whatever internet 
otocols?
(Given that we're talking DOS here; of course these services will be already 
running, in system space, on say an OS/2 or NT system).
Bink does that using its EXTMAIL feature; it just swaps itself out of the 
way, and launches the appropriate batch to run the external service.  Mine's 
simply triggered by receiving the string PPP during mailer handshaking; I 
actually did get it to run a KA9Q FTP server a time or two.  Now this sort of 
thing maybe could be added to Opus, and still leave a little room for Opus :)
 ME> I happen to work enough with HTML to construct pages and
 ME> have worked with
 ME> oec/mecca stuff from the beginning in opus.
 jb> I've also been doing HTML and CGI 12 hrs a day for just
 jb> about a year. It's a snap, where OEC is a real PITA and I
 jb> only do the minimal required to get done what I need.
You're still relying on a HTTP server to actually serve your HTTP client (ie, 
browser).  There's source code around for Web servers (Jigsaw, free with 100% 
Java source, as an interesting example) but we're not talking about any small 
project here; not even after you've got PPP/IP/TCP etc integrated and 
nning.
 ME> WEBBBS11.ZIP [    ] WebBBS runs on your Bulletin Board System
Is this Colin Wheat's gadget, that someone mentioned?
[..]
 jb> Install SZ, dial, go through door,  and switch to browser
 jb> mode. Straght PPP would be many X easier on the user,
 jb> though many X more difficult for the BBS author. I -do-
 jb> have an honest idea of how difficult it is to make it work.
:)  Multiply the number you first thought of by 10, and add 6 months.
 ME> ...... end of thread.
 jb> Too bad.
 jb> :-)
Ah, not always so easy .. :)
I don't mean to be a naysayer, because I think moving towards user browser 
connectivity is going to be vital to BBSes that want to survive, but I'm not 
at all sure that Opus can or ought to try including that, but maybe could 
just facilitate connections to whatever other protocols you want to run?  
O.
Cheers, Ian
---
---------------