Ok, thanks for the review and notes.

On 12/3/06, Alex Karasulu <[EMAIL PROTECTED]> wrote:

Hey Emmanuel,

A few notes on the meta schema.


Problems With Schema
--------------------

  o m-type-attribute-type is unknown to me.  I could not figure out what
you wanted to use it for.  I deleted it for now thinking we can redifine
it if need be.


This is obviously a mistake. You did well deleting this superfluous AT.

 o You made all attributes have a USAGE of dSAOperation and they were
set to NO-USER-MODIFICATION.  Schema changes can still be made by users
so these are userApplication attributes and are modifiable.


I don't think so for MetaAttributes. They should _never_ change.

Plugin Quirks
-------------

  o Unfortunately you must use lower case for strings 'objectclass' and
'attributetype'


ok, not a big deal. We may want to change that latter.

 o Order of attributes are important as defined strictly by RFC: i.e.
SINGLE-VALUE must go after SYNTAX but before NO-USER-MODIFICATION.


I don't think that we must respect the order, but however, we can stick to
it atm. I just think that this constraint (which is not explicit in the
RFCs) is what it is : a constraints to users. How many time did you had to
modify your schema file just because you have misplaced the SINGLE-VALUE
element? So in my mind, I think we should relax this constraint. The
important thing is that all the elements are present... But it can wait too.

General Rules
-------------

   o You must use OID values for SYNTAX even if you have to make up a
placeholder syntax for temporary use.


You are right...

  o By convention first character in names is lowercased and camel
humps are used.


yes, that's right.Let's keep the m-, but remove all the other '-' and  use
camel humps.

Here's the commit if you'd like to take a look and confirm correctness.

http://svn.apache.org/viewvc?view=rev&revision=481896

THanks,
Alex





--
Cordialement,
Emmanuel Lécharny

Reply via email to