FreeTDS Development Group

["James K. Lowden" <jklowden AT freetds.org>]

Medi Montaseri wrote:
> I modified bsqldb.c and provided a -w option (w for waitfor) such that
> -w 0   // default value, meaning don't throttle, blast away
> -w n  // where n is > 0 and it means sleep for n seconds
> -w -n  // where n is < 0 and it means sleep for ever

This is interesting.  Usually people are interested in going faster.... 
;-)

I think -wN could mean "wait N seconds" and -w0 could mean wait forever. 
There's no need for a "blast away" option; that's SOP.  

> In the course of changing bsqldb.c I noticed that the current design is
> reopen-ing stdin, stderr and stdout to support -o -e -i.
...
> 2- Since the case of -w -1 requires receiving input from stdin, it
> creates a conflict between -i which has been reopend as stdin. 
> 2.1 - I can either change the entire code and not use reopen but simply
> use a different descriptor for -i, -o, -e processing. Or
> 2.2 - I can temporarly reopen stdout and stdin while interacting with
> user and put it back to -o -i 

It seems to me it would be useful to be able to do this:

        echo $query | bsql -w0 ...

IOW, it would be nice to get user input regulating the statement
processing from somewhere other than stdin.  

But where?  I think getpass(3) is your friend.  It reads directly from the
terminal, bypassing stdin.  The input doesn't have to be a password, and
neither you nor the user cares what's typed except for the CR, so it
doesn't matter that it's not echoed.  

> Please let me know and  if it is a go...who should I coordinate this
> patch or checkin with ?

Best is to post a patch to Source Forge, second best to attach it to a
message posted here.  If your patch applied cleanly to the nightly
snapshot, it has the best chance of being included right away.  

--jkl