>
> > > > Varchar currently (in 6.2.1 and below) takes up the entire length specified
> > > > in the definition, despite the fact the value in it may actually be
> > > > shorter. Text takes only the space taken by the value.
> > >
> > > Thanks for the clarification. In this case, what happens with varchar's
> > > length if the original definition for that field leaves length undefined?
> > > Does it behave like text in that case?
> >
> > You really shouldn't be doing that. Not sure what happens. Not a good
> > idea:
> >
> > create table test (x varchar);
>
> ?? This was defined to be a varchar of unlimited length, much like, or identical
> to, text. Should this now be disallowed? If so, we can fix the parser to disallow
> it so people don't get misled.
Oh, I didn't know. There really is no difference between varchar with
no lenght, and text, but if it doesn't break anything, no problem.
>
> > > I also vaguely recall seeing a message last year about the use of indexes
> > > in queries: that in [some circumstances] indexes built on varchar fields
> > > don't get used and a sequential scan through all records takes place
> > > instead. Is there any distinction between varchar and text here?
> >
> > Don't remember that.
>
> This was probably Bruce's improvements to allow indices on some pattern matching.
> Doesn't make a distinction between these types in its behavior.
Not sure what to say on this. I remember that issue, but not how it
caused any problem.
--
Bruce Momjian | 830 Blythe Avenue
maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026
+ If your life is a hard drive, | (610) 353-9879(w)
+ Christ can be your backup. | (610) 853-3000(h)