Re: pgsql: Allow opclasses to provide tri-valued GIN consistent functions.

On 03/23/2014 06:07 PM, Greg Stark wrote:
> On Sun, Mar 23, 2014 at 1:22 PM, Heikki Linnakangas <hlinnakangas@vmware.com
>> wrote:
>
>> The reason for the typedef is precisely that an enum is not guaranteed to
>> be one byte. Tom suggested getting rid of the typedef, but it's needed to
>> make sure it's stored as one byte.
>>
>> I'll go add a comment to it, explaining why it's needed.
>
> If we're not using the enum type perhaps it would be better to make it a
> bunch of #defines? The main advantage of enum types is that the debugger
> knows what values are legal and can decode them for you.

Yeah, I think you're right. Changed it to use #defines, renamed the
typedef to GinTernaryValue as that's more descriptive, and added a brief
comment mentioning that a GinTernaryValue is compatible with a boolean.

- Heikki