Ned wrote:
>
>> My point is that if the purpose of the attribute exchange is to avoid
>> incompatibilities between sender and receiver, the current design isn't
>> going to work reliably. For example, there is the scenario of a client
>> sending a PS3 file to a PS2 IPP Printer.
>
>You're making the very big assumption that PostScript breaks down cleanly
by
>simple version numbers. But it never has been this way, and hasn't been
since
>Level 2 was defined. Devices are free to implement subsets of level 2 or
level
>3 and can check the declarations in a document to see if the document uses
>anything outside of the subset they implement. So a simple version number
just
>doesn't work very well for PostScript.
>
Wouldn't it be better than nothing? Can't we find something between blind
trial-and-error and requiring every client to have a priori knowledge of
every existing printer-make-and-model?
>However, the earlier example I saw was PCL. I don't know enough about PCL to
>say whether a simple version number is useful. If it is then defining a MIME
>version parameter for PCL might be appropriate.
>
>But while simple versioning through the use of MIME parameters is OK, anything
>more than that isn't something MIME parameters were designed to do. Nor are
>MIME content type parameters all that good at negotiating feature sets or
>anything similar.
>
>We have a mechanism for complex version and feature declaration and
>negotiation: Media feature tags.
>
But don't Media feature tags describe physical characteristics, rather than
how those characteristics are described in a PDL? Couldn't a given Mime
media feature have a different expression in different versions of PS or
PCL? [I admit I don't know much about media feature tags. Can you point
me to some references?]
>If attribute exchange is your goal you should be using media feature tags. MIME
>content type parameter information is only a small subset of the information
>available in this context. And if you need additional feature tags for
>PostScript or PCL, by all means define them.
>
The thing is, these would have to be generally accepted standards in order
to be useful for interop. It won't do any good for me to define my own new
tags.
> Ned
-Carl
This archive was generated by hypermail 2b29 : Mon Mar 12 2001 - 14:38:26 EST