Re: [Firebird-devel] Dmitry's "Future Development" paper

[Jim S. <ja...@ne...>]

Alex Peshkov wrote:

> Jim Starkey wrote:
>
>> Ian Newby wrote:
>>
>>> Hi Folks,
>>> Looks good. Only thing I miss is two new field types, boolean and 
>>> guid. These would make migration from other databases much easier.
>>>
>> After a long discussion, I'm not at all convinced that a GUID 
>> datatype makes any sense.  A GUID makes good sense as a domain / 
>> global field, but since there are only two operations defined for 
>> GUIDs -- generate and equality comparison -- I see no benefit for a 
>> GUID data type.
>>
>> Rather than introduce a limited boolean data type, I would much 
>> rather see general support for enumerated types.   An enumerated type 
>> can easily model the boolean values true or and false and lots more 
>> such as sex ("male", "female", "pre-marital", "extra-maritial", and 
>> "none"), prospective features ("good", "bad", and "ugly"), etc.
>>
>
> How do you plan to use boolean logic with enumerated types? What means 
> "male" XOR "female"?
>
The syntax would still have to be "sex = male or sex = female" (I 
haven't heard any requirement for XOR, but that's a different 
question).  This would also mean that a boolean reference would have to 
be "boolean_field = true" since the SQL language structure pretty much 
precludes use of a single identifier as a boolean expression.