IPP Mail Archive: IPP> Re: DIR - Revised text for SLP 'printer:' template - with updated comments

IPP> Re: DIR - Revised text for SLP 'printer:' template - with updated comments

Erik Guttman (Erik.Guttman@Sun.COM)
Tue, 10 Nov 1998 16:30:53 +0100

Tom, others,

I have a few comments on the latest draft.

> I'm not conversant with SLPv1 or SLPv2.
> What are the maximum lengths on STRING?
> In IPP we go to great pains to specify the maximum length
> of strings and to require that all implementations support
> the maximum length.

Strings have a maximum length of 2^16 in SLP. If you want
the strings to be compatible with use in IPP, by all means
state they have a maximum length in the description.

> name=STRING
...
> description=STRING
...
> version=0.2
...
> url-syntax=

These fields have been renamed in the latest draft of the service scheme
document to template-name, template-description and template-url-syntax.
The service scheme document is currently in IESG last call (so it won't
change much any more. It should go to proposed standard soon.) "url-syntax"
and "version" are already in the template. These attributes need to be
renamed. Entries for "template-name=service:ipp" and
"template-description=<text>" need to be added. Note that the latter
attribute is distinct from the "description" attribute in the template
which describes the given printer. The template-description attribute
is a description of the *template*.

These attributes are used by slp implementations which process templates.
Every template has its version, service type name, a description and a
URL syntax section irrespective of whatever other information it contains.

> See mail sent on 11/6/1998 proposing to use the > character to
> separate ordered value within a single-valued string.

Please note that the '>' character is reserved, so it would have
to be escaped in the template text. Thus, if you were to include

"tls > ssl3 > none"

as the value of the uri-security-supported attribute, you would
have to encode this as "tls \3E ssl3 \3E none" when transmitted
in SLP messages. Note that this should be transparent to applications
using SLP and to (for example) a directory service where the attribute
value was stored.

> uri-supported = STRING L M O
> # IPP 'printer-uri-supported'
> # The ordered list of URI supported by this printer,
> # correlated with the 'uri-security-supported' attribute.

It would be very helpful if this discussed exactly what an
ordered list is. For instance, it is a

1*DIGIT SP <value>

Where the initial number is the ordinal value. So "1 foo"
precedes "2 bar" in terms of order, whatever sequence these
values might be transmitted in.

> uri-security-supported = STRING L O
...
> # Each security value is separated by greater-than ('>')

Does this indicate the ordinal value? Note that if you
have multiple values concatenated into one using '>' SLP
will consider them to be one value, not several. For example:

"A \3E B \3E C" (which is SLP encoding for 'A > B > C')

would be considered one value. "A,B,C" is three values, but
without any ordering implied.

> alternate>>>> See the mail sent 11/06/98 to make this ordered
> values as a single valued string with each value separated by
> greater-than ('>').

Why do you want to give up the fact that they are distinct
values? Why not include an explicit ordinal as we discussed
on the list and I suggested above?

> Is the -1 convention something that is used in SLP protocol?

SLP interprets "-1" as an integer with the value minus one.

------------------------------------------------------------------
Erik