Jay writes:
>>> JK Martin <jkm at underscore.com> 05/05 3:26 PM >>>
> As far as deriving base constraints goes, to me the real question is:
>> Must an "IPP Printer" must be able spool print jobs?
>> I would like to see an answer of "Yes", but clearly this is where desktop
> printers (and many, if not most, network printers) fail to perform. And I
> believe this is usually the reasons behind most case of "we can't do that
> due to printer limitations."
I fail to see why a "yes" answer matters? If someone decides to implement
IPP in a printer that can not spool, what changes? If it is too busy
processing one job and another comes in from another client, it returns a
"busy" or a "out of spool space" just as in the case where a server
implementation supports 2000 spooled jobs and the 2001st job comes in. When
it is asked how many "queued jobs" it answers "0" or "1" depending if it is
processing a job or not. If it is too busy to even establish a client
connection (it only supports one at a time) then the second one times out,
just as if it supported 500 client connection and the 501st comes in and
fails and experiences a time out.
Will anybody implement IPP in a printer like this? Probably not. Does the
model preclude them from doing so? No. Will this type of printer ever be
used as as a shared network printer? Probaly not, but if it is, and it does
not embed IPP then it can be front ended by a "server" that does.
> And I
> believe this is usually the reasons behind most case of "we can't do that
> due to printer limitations."
I believe the only time we should change the model or make decisions based
on info like "we can't do that" is when we are talking about what exists in
reality today and the near future - like "many printers that support a PDL
like PostScript can't support overriding PDL instrcutions with exteranal
attributes" In view of that statement, we must "work" in the exising world
and not just some future fantasy world. I don't want to succumb to the
argument that since the printing spectrum is soooo wide, we can not do one
thing or the other since it fails at the exteme ends. IPP should fall
comfortably in the middle of the spectrum and all for niche extensions to
support the fringes of the spectrum.
Scott
Scott