Re: syntaxes-09 notes

At 04:26 AM 3/8/2005, Hallvard B Furuseth wrote:
>Kurt D. Zeilenga writes:
>>At 12:56 PM 3/6/2005, Hallvard B Furuseth wrote:
>>>Kurt D. Zeilenga writes:
>>>>>>>3.3.11.  Facsimile Telephone Number
>>>> The E.123 format is best considered a free format.  While it
>>>> can certainly be argued that a more precise format would be
>>>> better, it is not generally within our scope to update X.500
>>>> syntaxes (or to introduce new syntaxes).
>>>
>>> That's fine by me.
>>>
>>> However, you rejected a complaint that OpenLDAP did not accept
>>> "---" as a phone number.  Have you changed your mind, or do I
>>> misunderstand what you mean by free format?
>>
>> E.123 a free format for representing a telephone number.  I
>> argue that it is not a telephone number and, much like dates
>> like 31 February 2006, can be rejected by the implementation.
>
>Then [Syntaxes] should at least allow invalid telephone numbers to be
>rejected.  An in standards context, I'm not sure what "not a valid
>telephone number" could mean other than "does not follow E.123".

"telephone number" is an abstraction.  E.123 provides a format
for representing values of that abstraction.  Aside from the
input violates the format, input can violate the abstraction.
This, IMO, is a general principle.  Servers are not required
to understand details of the abstraction, but if they are, they
certainly can reject values which are not consistent with
that abstraction.

>I would note that we've been stuffing phone numbers from text fields in
>other databases into LDAP's telephoneNumber, and while these textual
>phone numbers may contain enough info or context to make them clear to a
>human, it may not be possible for a computer program to decide what they
>mean.

I also note that the telephoneNumber abstract is actually an
"international telephone number".  "52813/52828" can be consider
invalid because it represents multiple numbers.  "52813/52828"
can be considered invalid because these do not represent
international telephone numbers.

>E.g. "52813/52828" = +47-228 52813 and +47-228 52828 (+427-228*
>belongs to our organization).


>Or deciding whether something is a phone
>number + trailing line number, or an international number where someone
>has omitted a '00 or '+' country prefix, or two 5-digit phone numbers.

The program can reject values that don't start with "+" country
prefix (ignoring other characters that can be ignored).  The
abstraction is, after all, an international telephone number.

>The data sources have been cleaned up considerably now, but some years
>ago too restrictive phone number syntax would have meant we had to omit
>a lot of numbers from the directory.

Or, to look it another way, the less restrictive syntax allows
a lot of garbage into the directories.

>Whether such a situation is
>LDAPbis' problem or the organization's problem is another matter, of
>course.