perl-ldap-dev

Reguested Schema data

[Clif H. <ch...@po...>]

Graham,  

I have attached 2 files that contain data about the schema 
problem.  

ldap.oids.gz contains oids that do not return data.  

ldap.schema.gz is a complete schema from one of my test 
netscape directory servers.  

If possible, possibly now or in the future, I would like to 
see us do away with changing the case of the oids and names. 
I have not had a chance to see how hard this would be.  

This would do away with some confusion and I know in my case would do
away with creating multi-valued oid strings.  On one of my servers I
have an objectclass named mhsMessagingStore and an attribute aliased as
MHSMessagingStore, when we lower case the names and aliases in this case
we create the same name pointing to 2 different
oids.                                                                                 

I might be able to look into this problem some this weekend.

Regards,

Clif Harden          ch...@po...

Re: Reguested Schema data

[Graham B. <gb...@po...>]

On Thu, Jan 03, 2002 at 09:50:04PM -0600, Clif Harden wrote:
> Graham,  
> 
> I have attached 2 files that contain data about the schema 
> problem.  
> 
> ldap.oids.gz contains oids that do not return data.  
> 
> ldap.schema.gz is a complete schema from one of my test 
> netscape directory servers.  

Thanks.

> If possible, possibly now or in the future, I would like to 
> see us do away with changing the case of the oids and names. 
> I have not had a chance to see how hard this would be.  

You cannot. Attribute names etc are not case sensetive. So the module
must ignore case when doing a lookup for the name supplied.

Graham.

> 
> This would do away with some confusion and I know in my case would do
> away with creating multi-valued oid strings.  On one of my servers I
> have an objectclass named mhsMessagingStore and an attribute aliased as
> MHSMessagingStore, when we lower case the names and aliases in this case
> we create the same name pointing to 2 different
> oids.                                                                                 
> 
> I might be able to look into this problem some this weekend.
> 
> Regards,
> 
> Clif Harden          ch...@po...

Re: Reguested Schema data

[Chris R. <chr...@me...>]

Graham Barr <gb...@po...> wrote:
> On Thu, Jan 03, 2002 at 09:50:04PM -0600, Clif Harden wrote:
>> Graham,  
>> 
>> I have attached 2 files that contain data about the schema 
>> problem.  
>> 
>> ldap.oids.gz contains oids that do not return data.  
>> 
>> ldap.schema.gz is a complete schema from one of my test 
>> netscape directory servers.  
> 
> Thanks.
> 
>> If possible, possibly now or in the future, I would like to 
>> see us do away with changing the case of the oids and names. 
>> I have not had a chance to see how hard this would be.  
> 
> You cannot. Attribute names etc are not case sensetive. So the module
> must ignore case when doing a lookup for the name supplied.
> 
> Graham.

Correct, however Clif also adds:

>> away with creating multi-valued oid strings.  On one of my servers I
>> have an objectclass named mhsMessagingStore and an attribute aliased as
>> MHSMessagingStore, when we lower case the names and aliases in this case
>> we create the same name pointing to 2 different
>> oids.

This means that we need multiple oid 'namespaces', as Clif has two OIDs -
one an objectclass and one an attribute - with the same name.

We currently have a single %oid object per Schema object, so we could do
something like prefix each oid string with "at" or "oc". I'm not sure if
that's the best solution, or if this would cause other problems. (For
example, what would you do with values for an attribute that has an OID
syntax - would the values be 'at' or 'oc' OIDs?)

Cheers,

Chris

Re: Reguested Schema data

[Graham B. <gb...@po...>]

On Mon, Jan 07, 2002 at 12:53:42PM -0000, Chris Ridd wrote:
> > You cannot. Attribute names etc are not case sensetive. So the module
> > must ignore case when doing a lookup for the name supplied.
> > 
> > Graham.
> 
> Correct, however Clif also adds:
> 
> >> away with creating multi-valued oid strings.  On one of my servers I
> >> have an objectclass named mhsMessagingStore and an attribute aliased as
> >> MHSMessagingStore, when we lower case the names and aliases in this case
> >> we create the same name pointing to 2 different
> >> oids.
> 
> This means that we need multiple oid 'namespaces', as Clif has two OIDs -
> one an objectclass and one an attribute - with the same name.
> 
> We currently have a single %oid object per Schema object, so we could do
> something like prefix each oid string with "at" or "oc". I'm not sure if
> that's the best solution, or if this would cause other problems. (For
> example, what would you do with values for an attribute that has an OID
> syntax - would the values be 'at' or 'oc' OIDs?)

This is inline whith what I have been thinking. Not doing the lc() is not the solution.

Internally we need to split %oid into several say $oid{at}{$oid}

That would also allow us to get all attributes easily, which we currently do
differently.

We then need to look at the API. Currently we do $schema->item($what), I think
the API needs a way to specifiy, get data from this attribute etc. as well as
the general get me this.

I have a few ideas in my head. I will see if I can put something together.

Graham.

Re: Reguested Schema data

[Clif H. <cl...@di...>]

Chris,

I sent Graham a new Schema.pm file last Friday night that addresses
the problems I was seeing with missing data and the 
attribute/objectclass naming situation.

Graham is reviewing my changes.

Later,

Clif


> 
> Graham Barr <gb...@po...> wrote:
> > On Thu, Jan 03, 2002 at 09:50:04PM -0600, Clif Harden wrote:
> >> Graham,  
> >> 
> >> I have attached 2 files that contain data about the schema 
> >> problem.  
> >> 
> >> ldap.oids.gz contains oids that do not return data.  
> >> 
> >> ldap.schema.gz is a complete schema from one of my test 
> >> netscape directory servers.  
> > 
> > Thanks.
> > 
> >> If possible, possibly now or in the future, I would like to 
> >> see us do away with changing the case of the oids and names. 
> >> I have not had a chance to see how hard this would be.  
> > 
> > You cannot. Attribute names etc are not case sensetive. So the module
> > must ignore case when doing a lookup for the name supplied.
> > 
> > Graham.
> 
> Correct, however Clif also adds:
> 
> >> away with creating multi-valued oid strings.  On one of my servers I
> >> have an objectclass named mhsMessagingStore and an attribute aliased as
> >> MHSMessagingStore, when we lower case the names and aliases in this case
> >> we create the same name pointing to 2 different
> >> oids.
> 
> This means that we need multiple oid 'namespaces', as Clif has two OIDs -
> one an objectclass and one an attribute - with the same name.
> 
> We currently have a single %oid object per Schema object, so we could do
> something like prefix each oid string with "at" or "oc". I'm not sure if
> that's the best solution, or if this would cause other problems. (For
> example, what would you do with values for an attribute that has an OID
> syntax - would the values be 'at' or 'oc' OIDs?)
> 
> Cheers,
> 
> Chris
> 
>