FreeTDS Development Group

["Jonas Benjaminsson" <jonas.benjaminsson AT infogate.se>]

:: As Brian said, it's best to post your "diff -ru" output to the sourceforge
:: patches page.  If you've got CVS functioning, it's a good idea to keep your
:: tree up to date.  Just let "cvs update" merge changes to the base into your
:: workfiles.  That way, your patch can be cleanly applied when you send it,
:: and you won't wind up fixing something that becomes obsolete.  

I will do that.

:: Slow is better than broken.  
:: 
:: Speed is not an issue here, really, because the client CPU is so much 
faster
:: than data transfer to the database.  In any case, writing the null device
:: sure oughta be fast, and any algorithm that could determine the right
:: allocation size would have to do something a lot like what vfprintf(3) 
does.

I agree.

:: 
:: 
:: Wouldn't this be portable?:
:: 
::      #ifdef WIN32
::      #define TDS_NULL "NUL:"
::      #else
::      #define TDS_NULL "/dev/null"
::      #endif
::      ...
::      if ((fp = fopen(TDS_NULL, "w")) == NULL)

You're so right, it sure did work.

:: Besides, Win32 OS's ship with a vfprintf(), so you don't really have to
:: worry about compiling or using the replacement.  

Yes it does, and that's what I'm going to use.
Just thought the aproach to write to a file was somewhat odd.

:: Lucka till med din project, and regards, 

Tack!  :-)


/Jonas