DR> Which brings up a point, should we go DTD or schema? I think schema
 DR> but I am no expert in these matters.

Me neither. Schema is more advanced so I think that's what we need to use.
For one thing it allows us to lock down the _type_ of the data fields
(integer, string, date) and even valid formats (specified by a regexp of
some sort if I'm not mistaken). But I think we can worry about that
later... :-)

 JSpstn>
 JSnumber/number>
 JSflagflag>
 JSflagflag>
 JSspeed/speed>
 JS/pstn>

 DR> hmm... so you prefer children to attributes?

I dunno... Decisions decisions... :-)

We can probably worry about that later too, when we know more about what we
want to stuff in there...

 DR> For the flags, wouldn't
 DR> it be best to go a step further with the type of flag? At least for the
 DR> known
 DR> types and then have a generic flag type?

 DRflagSpeed/flagSpeed>
 DRflagOnline/flagOnline>
 DRflagModem/flagModem>
 DRflagUpdate/flagUpdate>
 DRflagCompression/flagCompression>

You're right. But that makes "flag" a bit redundant so make those
"speed", "online" etc. instead. BTW, by
"modem" you mean modem protocols right? Like V34 and V90? How
about "protocol" instead?

 DR> Ah yes, I forgot about ISDN in my proposal. Again, shouldn't the
 DR> various flags be defined by name a little more?

Yes...

 JS>> Here "pstn", "isdn", "inet" and
"transx" are different "transports",
 JS>> which need different sets of info to be used. E.g. internet
 JS>> transports
 JS>> don't have phone numbers, and modem transports don't have IP info.
 JS>> This
 JS>> needs a lot of thinking. Anyone have any good ideas?

 DR> Yes, I think you are leaving out transport methods. For Inet you are
 DR> using
 DR> ONLY BinkP while there is also TELNET, IFCIO, VMODEM, FTP etc.

Yes of course. I was just making some examples :)

 DR> I think that Transx should also fall under INET.

Yes, technically it would, but I was thinking along the lines of how one
actually use them. Binkp, Telnet, VMP & Ifcico are direct
"protocols" running over TCP. Transx is an other
"type", since it doesn't use connections but simply emails the
packet to an email address... You don't use IP addresses with Transx. Make
any sence?

 DR> I think we should look at it this way. If an element can possibly have
 DR> an attribute then it should be a child. If it cannot have an attribute,
 DR> then /IT/ should be an attribute. If it makes it easier for parsing then
 DR> perhaps everything could be a child.

Let's keep it open for now. :-)

  Jesper,
  yeppe{at}enjoy.cc
---