Re: Logical Implication - Mailing list pgsql-hackers

From Zsolt Parragi
Subject Re: Logical Implication
Date
Msg-id CAN4CZFPUmkNyYP=MJbCE3OC21-De9AYhJXOD_wd8Q06E6xBzqQ@mail.gmail.com
Whole thread
In response to Re: Logical Implication  (Vik Fearing <vik@postgresfriends.org>)
Responses Re: Logical Implication
List pgsql-hackers
Hello!

One test nitpick:

+SELECT istrue IMPLIES isnul IS NULL FROM booltbl4;

istrue IMPLIES (isnul IS NULL) and (istrue IMPLIES isnul) IS NULL are
both true. isfalse wouldn't have the same issue.



 %left        INTERSECT
+%right        IMPLIES
 %left        OR


And +1 for using %nonassoc instead

That way "a IMPLIES b IMPLIES c" would be a syntax error, leaving open
the later possibility of making it right associative. The other way of
going with right, and then restricting it later doesn't seem that
clean, even with documenting it, as right associative use can appear
as text in plpgsql bodies and similar.



pgsql-hackers by date:

Previous
From: Alexander Lakhin
Date:
Subject: Re: REPACK (CONCURRENTLY) can crash a logical decoding session
Next
From: Andrew Kane
Date:
Subject: Re: Fix conversion warnings in headers