AW: [HACKERS] DROPping tables with SERIALs - Mailing list pgsql-hackers

From Zeugswetter Andreas IZ5
Subject AW: [HACKERS] DROPping tables with SERIALs
Date
Msg-id 219F68D65015D011A8E000006F8590C60180FAB1@sdexcsrv1.sd.spardat.at
Whole thread Raw
List pgsql-hackers
>> >> No, I think it should *only* substitute for NULL.  Why assume>> >> zero is special?>> >> > As I remember this
ishow SERIAL works in Informix. >> >> Ah.  OK, if that's what they do then I agree we ought to act the
 
same.>I hope that this wasn't to say that the SERIAL should substitute
for 0>as well as NULL.  That would be quite annoying.  I say stick to
NULL.
Yes,  0 is an allowed value for a serial, that should not be
substituted.I like the behavior that a NULL is substituted.Informix is a little dull here. It forces a not null
constraint,and
 
will thereforeraise an error if a null is inserted. I don't see any advantage in
this behavior.(You are forced to skip the field in the insert statement to get a
generated serial)
Andreas



pgsql-hackers by date:

Previous
From: "Jackson, DeJuan"
Date:
Subject: RE: [HACKERS] DROPping tables with SERIALs
Next
From: Tom Lane
Date:
Subject: Re: [HACKERS] Two pg_dump ugly bugs :-(