FreeTDS Development Group

["Lowden, James K" <LowdenJK AT bernstein.com>]

> From: Bill Thompson [mailto:thompbil AT exchange.uk.ml.com]
> Sent: June 14, 2002 11:19 AM
> 
> > Personally, I've never found the format of the error file 
> very easy to
> > follow.  Do you think we have to adhere to it?  I suppose 
> people have
> > scripts that parse the error file.  I wonder whether it's 
> more important to
> > do what Sybase does, or to try to do better.  
> > 
> 
> Hmmm. I haven't really looked into this yet, although I intend to.
> If it looks easyish to do, I'll probably stick with the 
> Sybase format, for the reasons you've outlined. 

OK.  I have another issue for you to consider, too, while the design is
young.  

bcp has three output streams: stdout (n rows copied), stderr, and -e.  An
unfortunate choice on the part of Sybase was that unless -e was set, no data
parsing error feedback was offered.  You'd see just "error parsing line 6"
or somesuch, but no details.  

I think it would be better (and more free-software-like, FWIW) to just splat
the would-be -e error file data to stderr in the event there's an error and
no -e filename has been set.  

The reason I mention this is that it implies that the -e filehandle is a dup
(?) of stderr unless a -e filename is given and successfully opened.  

> BTW User guide follows by 
> separate email to you...

thanks++; I'm wrestling with the UG, and at the moment it's not clear who's
winning.  We'll know on Monday.  

--jkl