Dave Page wrote:
>
>
>
>>-----Original Message-----
>>From: Andreas Pflug [mailto:pgadmin@pse-consulting.de]
>>Sent: 26 July 2004 18:51
>>To: Dave Page
>>Cc: Vitaly Belman; pgadmin-hackers@postgresql.org
>>Subject: Re: [pgadmin-hackers] [pgadmin-support] Schemas
>>causing problems :(
>>
>>Dave Page wrote:
>>
>>
>>>Perhaps the search needs to be more clever though. Consider the
>>>following:
>>>
>>>search_path: public
>>>custom type: public.text
>>>
>>>In this case we might need to specify pg_catalog.text to
>>
>>get the right
>>
>>>one.
>>
>>This is getting a nightmare...
>
>
> Yup. So where are we with this? I'm thinking that for any object, we
> check each schema in the search path in turn (and then pg_catalog if
> it's not explicitly included) for an object of the same type with the
> same name (noting that sequences, tables and indexes are all effecitvely
> the same object type in this situation). If we get to the parent schema
> without first finding another matching object in a previous schema, then
> and only then do we suppress the schema name.
>
> How does that sound?
Complicated.
Before trying to suppress the schema, I'm checking for any duplicate
type names now. If so, schema is always emitted. In your sample:
pg_catalog.text and public.text.
Regards,
Andreas