| From: | Mike Nolan <nolan(at)gw(dot)tssi(dot)com> | 
|---|---|
| To: | peter_e(at)gmx(dot)net (Peter Eisentraut) | 
| Cc: | cool_screen_name90001(at)yahoo(dot)com (CSN), pgsql-general(at)postgresql(dot)org | 
| Subject: | Re: Sorting by numerical order | 
| Date: | 2004-03-14 22:59:45 | 
| Message-ID: | 200403142259.i2EMxk1F007960@gw.tssi.com | 
| Views: | Whole Thread | Raw Message | Download mbox | Resend email | 
| Thread: | |
| Lists: | pgsql-general | 
> CSN wrote:
> > Is it possible to have items sorted like:
> >
> > abc 1
> > abc 2
> > abc 10
> > abc 20
> 
> Cast them to a numeric type.
You can't cast a variable char type to a numeric, though apparently
you can cast text to numeric.  (I'd be curious to know why one but not
the other.)
However, if I understood the original question, 'abc 1' and 'abc 20' are 
contents of a single column, which might make separating out just the
numeric part tricky.
On a somewhat related subject, how difficult would it be to write an
'isnumeric' function as part of the source code as opposed to a user
function in pgsql or perl?
I have some data that is mostly numeric but not always so.  I know from 
other columns whether a particular row is supposed to contain a numeric 
value or not, but the data is coming from a legacy environment
and isn't 100% reliable.  
An 'isnumeric(column)' boolean function in a case statment would be 
very helpful so that I can program around the bad data.
I'll probably write it in perl, but would making it part of the source
code provide significantly faster execution?
--
Mike Nolan
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Eric B.Ridge | 2004-03-14 23:06:38 | Re: How reliable are the stats collector stats? | 
| Previous Message | Frank Finner | 2004-03-14 22:06:53 | Re: PostgeSQL problem (server crashed?) |