Roland Roberts <roberts(at)panix(dot)com> writes:
>>>>>> "Tom" == Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> writes:
Tom> Must have been quite a few versions back; attlen has been -1
Tom> for variable-length datatypes for as long as I've been paying
Tom> attention. (Of course char(n) isn't *really* variable length,
Tom> but it's treated that way so that the representation is the
Tom> same as for varchar(n) and text.)
> When did this change? I haven't looked 6.5, but I thought 6.3 still
> has bpchar as fixed length and char(n) was blank padded whereas
> varchar was not.
char(n) was and is blank-padded, varchar(n) was and is not. Sorry
if I was imprecise enough to convey a different impression.
I'm fairly sure that attlen had its present semantics in 6.4. 6.3
was before my time; any old-timers care to comment?
regards, tom lane
In response to
pgsql-bugs by date
|Next:||From: Armand Delcros||Date: 2000-05-04 11:00:33|
|Subject: [RC2]:Compilation failure solaris 2.7 stringinfo.c|
|Previous:||From: Roland Roberts||Date: 2000-05-04 03:26:56|
|Subject: Re: Need some help: attlen is pg_attributes gives a negative value.. .|