Re: short names (descriptors) proposal

[Michael Ströder <michael@stroeder.com>]

(Note: corrected citation)

Ramsay, Ron wrote:
Kurt Zeilenga wrote:
>>
>> At 06:00 AM 2002-11-04, Michael Ströder wrote:
>>
>>>Timothy Hahn wrote:
>>>
>>>>Regarding the use of the same name for an attribute type and an object
>>>>class - I agree that it should be "discouraged".  I don't think the
>>>>specifications need to explicitly DIS-allow it.
>>>
>>> Maybe it would be helpful for application developers if the standard
>>> document explicitly states that an attribute type and an object class
>>> MAY
>>> have the same name although discouraged. I had to learn it the hard
>>> way...
>>
>> How about adding the following to 6.2:
>>   Note: While discouraged, one name may (for historical reasons)
>>   be an alias for multiple OIDs each referring to a different kind
>>   of schema element.  That is, there may be an objectClass 'fubar'
>>   and an attributeType 'fubar'.
>
> This phrasing, and the previous one, seem to suggest it is a really good
> idea to use the same name in different namespaces.

I agree that my phrasing with MAY was inappropriate and could encourage 
people to do so.

> Why not
>
> The same name SHOULD NOT be used for different OIDs, even if these OIDs
> refer to different schema elements.

I agree the SHOULD NOT sounds stronger than "While discouraged..".

But I can argue the other way round for your phrase: Many naive implementors 
could read this phrase above like it never happens. It was my original 
intention to give an explicit warning.

How about this:

The same name SHOULD NOT be used for different OIDs. But implementations 
MUST be prepared that the same name might be used for multiple OIDs each 
referring to a different kind of schema element.

Ciao, Michael.