At 09:13 PM 3/11/2003 -0600, Craig Leonardi wrote:

>Hi Helen,
>
>Thanks for the quick reply. Comments embedded below.
>
> >
> > Could the "3-4 hours" mean that nobody calls a UDF until then?
> >
>Possibly. Several functions are used from two libraries:
> ib_udf.so
> udflib
>I'm trying to touch each procedure and view that calls UDF's but have been
>unable to reproduce any UDF-related error condition. Sigh.
>
> >
> > Have you studied the release notes to see the security changes affecting
> > UDF access and set things up accordingly on the server?
> >
>No. I'm looking at these things as you suggest. Thanks.
>
>One thing that still bothers me though is the file name. I can't find
>'libudflib.so' anywhere. It's not part of the firebird.msg file so I don't
>think the error message is misleading. I suspect that somewhere in FB 1.5,
>the file libudflib.so is getting called.

It will relate to the module name as you have it declared in
rdb$functions. This UDF library isn't known to Firebird at all, so there's
no reason for it to be in the .msg file. In searching for the module, my
guess it's going through a sequence of educated guesses. E.g. if you have
the module declared as "udflib" one of its tries might be to prepend "lib"
and add a ".so" suffix.

Did you have this database on Linux all along? Have you put the Linux
version somewhere? If you have, it should be suffiicient to put a symlink
from it to the UDF directory, naming it libudflib.so.

>Another potential issue is udflib. As you know, this was originally created
>by Shieck (of censorship fame). Do you know if this library is compatible
>with FB 1.5? It appears to be when using IB_SQL and isql. But, single
>statements are obviously not the same as a production environment.

No idea. But if you have a Linux version that works with the tools, and
the 1.5 server, I can't see it not working in all environments.

regards,
Helen