directory-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefan Seelmann <seelm...@apache.org>
Subject Re: [jira] Created: (DIRSTUDIO-109) Schema Entry Deletes Must be Sequenced
Date Wed, 23 May 2007 20:42:07 GMT
Ole,

I don't think it makes sense to add such semantics into the default
delete operation of the browser.

What we could do is the following: PAM and I would like to extend the
current Schemas Editor to operate directly on the schema partition of
ADS instead of .schema files. So we have to add this functionality into
the Schemas Editor:

- When saving a new schema into the schema partition we have to consider
the correct order (syntaxes before matching rules before attribute types
before object classes)

- When deleting a schema we have to consider the inverse order

Sounds good?

Regards,
Stefan Seelmann



Ole Ersoy schrieb:
> 
> 
> Alex Karasulu wrote:
>> Ole,
>>
>> This is not a bug.
> 
> Yes - I know - Sorry - I categorized
> it as a bug initially and then when I got
> done documenting it, I realized that it
> is really more of a new feature, so I added
> the "Disclaimer" at the top of the description.
>>
>> The schema subsystem is just telling you that you cannot delete a
>> syntax that is currently in use by
>> an attribute.  
> 
> Right.  That's why it has to delete
> the attributeType entry first,
> then the syntax entries.
> 
> The key is that I'm deleting the parent of both
> the syntax entries and attributeType
> entries, so LS can decide the sequence
> of the deletes.
> 
> You want cascade deletes? This will not work because you
>> can have more than one
>> attribute that depends on this syntax.
> 
> Just remember that we are
> deleting the parent of both syntaxes
> and attributeType entries in this case.
> 
> So we have to delete
> all the attributeType entries first,
> then we can delete the syntaxes.
> 
> The same is true if we have objectClasses
> that depend on the attributeType entries.
> 
> We have to delete the objectClasses,
> then the attributeTypes, then
> the syntaxes, and finally the
> container for all these entries.
> 
> Now if there are syntax entries
> in the server that also depend
> on these, then we'd have to delete
> these as well.
> 
> So this new feature would save
> save time when deleting schema
> containers that are not being
> used by "real" entries.
> 
> Cheers,
> - Ole
> 
> 
> 
> SNIP


Mime
View raw message