FreeTDS Development Group

[Frediano Ziglio <freddy77 AT gmail.com>]

Il giorno gio, 09/07/2009 alle 17.08 -0400, James K. Lowden ha scritto:
> Cedric ROUVRAIS wrote:
> > > Since freebcp supports the -b batchsize option, should the
> > > rows_written be compared to the const 1000 or to pdata->batchsize ?
> > 
> > Sybase BCP does not take into consideration the batch size for printing
> > to stdout progress. In other words no matter what value you put for the
> > -b option, it will always send to stdout every 1000 rows the message
> > "x000 rows successfully bulk-copied to host-file.".
> 
> I have always found that to be a misfeature.  You extract a 1,000,000 rows
> and get 1,000 lines of "1000 rows written to host file".  
> 
> I prefer the symmetry of mating the value of -b to the feedback in both
> directions.  On uploading, of course, -b controls the commit batch size
> (and feedback on standard output).  On extraction there's no commit, so
> its only effect is to reduce the noise on the terminal.  
> 
> Thank you for the code.  I hope to look at it this weekend unless Frediano
> beats me to it.  
> 

Personally I looked at bcp.c code. I think it would be better to change
len passed to dbconvert in order to avoid trimming. We use a len -1 or
-2 so dbconvert trim but passing a specific len cause dbconvert not to
trim!

Frediano