FreeTDS Development Group

[Harry Felder <harry AT rentec.com>]

Hi again:

Sorry, I got distracted and never generated a patch to fix the below discussed
problem with a zero-length password.

The problem occurs with unixODBC's isql in a call like:

        isql DSN UID ''

With the patch I can connnect fine. The patch is against the
latest cvs.

Also, SQSH 2.1 doesn't seem to compile at the moment since it includes
tds.h which includes config.h, which isn't installed. Should I configure it
it to be looking directly in the private freetds build include directory
instead of the public directory for installed include files?

thx,

hf



ROE: diff -c odbc.c.ORIG odbc.c
*** odbc.c.ORIG Fri Aug  9 14:05:15 2002
--- odbc.c      Thu Aug 15 14:01:22 2002
***************
*** 695,701 ****
      }

      /* password */
!     if ( szAuthStr && (*szAuthStr) )
          strcpy( szPassword, szAuthStr );
      else
      {
--- 695,701 ----
      }

      /* password */
!     if ( szAuthStr)
          strcpy( szPassword, szAuthStr );
      else
      {


On June 17, 2002, Brian Bruns wrote:
>
> I can't say 100% for the current code, but I know this was possible in the 
> past and I'm not familiar with any changes in that area.  What are you 
> trying to connect with? It could be a application issue.  I've been able 
> to do it with SQSH.
>
> Brian
>
>
> On Mon, 17 Jun 2002, Harry Felder wrote:
>
>
> > Hi all:
> >
> > I have a development MSSQL DB which was set up with a zero-length password,
> > which apparently MSSQL is quite happy to accept. Our MSSQL wizard says he has
> > lots of experience w/ setting up throw-away development DB systems this way.
> >
> > It appears that (the current cvs of) freetds prevents using a zero-length
> > password to connect, although commenting out the test seems to allow
> > the connection to succeed.
> >
> > So, is the check really unnecessary or is there more to this than meets my
> > inexperienced eye? (Yes, I know the dangers of such a "password"--I've got
> > lots of experience in those issues!)
> >
> > If the check can be removed, I can generate a patch to that effect.
> >
> > thx,
> >
> > hf