2013/1/7 Robert Haas <robertmhaas@gmail.com>:
> On Mon, Jan 7, 2013 at 2:14 PM, Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
>> Kohei KaiGai escribió:
>>
>>> Function and collation are candidates of this special case handling;
>>> here are just two kinds of object.
>>>
>>> Another idea is to add a function-pointer as argument of
>>> AlterNamespace_internal for (upcoming) object classes that takes
>>> special handling for detection of name collision.
>>> My personal preference is the later one, rather than hardwired
>>> special case handling.
>>> However, it may be too elaborate to handle just two exceptions.
>>
>> I think this idea is fine. Pass a function pointer which is only
>> not-NULL for the two exceptional cases; the code should have an Assert
>> that either the function pointer is passed, or there is a nameCacheId to
>> use. That way, the object types we already handle in the simpler way do
>> not get any more complicated than they are today, and we're not forced
>> to create useless callbacks for objects were the lookup is trivial. The
>> function pointer should return boolean, true when the function/collation
>> is already in the given schema; that way, the message wording is only
>> present in AlterObjectNamespace_internal.
>
> It seems overly complex to me. What's wrong with putting special-case
> logic directly into the function? That seems cleaner and easier to
> understand, and there's no real downside AFAICS. We have similar
> special cases elsewhere; the code can't be simpler than the actual
> logic.
>
Does it make sense an idea to invoke AlterFunctionNamespace_oid()
or AlterCollationNamespace_oid() from AlterObjectNamespace_internal()
for checks of namespace conflicts?
It can handle special cases with keeping modularity between common
and specific parts. Let's consider function pointer when we have mode
than 5 object classes that needs special treatment.
Thanks,
--
KaiGai Kohei <kaigai@kaigai.gr.jp>