At 04:57 PM 31/03/2004 +0200, you wrote:

> > Ummm....how do I answer this question? :-)
> >
> > I am saying that the script does stop executing (fails) at the point of the
> > error. If autoddl is on or commits were called on previous statements, the
> > previous statements would "stick"; but the failed statement and all after
> > it would not be executed, i.e. the script stops executing. (I hope there
> > are no single-double-triple negatives in that paragraph. <g>)
>
>Try this script:
>
>CREATE DATABASE 'c:\test.fdb' USER 'SYSDBA' PASSWORD 'masterkey';
>CREATE TABLE T (i INTEGER);
>INSERT INTO T VALUES (10);
> /* Although this causes "arithmetic exception, numeric overflow, or
> string truncation" ... */
>INSERT INTO T VALUES (10/0);
>INSERT INTO T VALUES (20); /* ... this statement will be executed ... */
>SELECT * FROM T; /* ... as you can verify by this SELECT, ... */
>DROP DATABASE; /* ... and the database will be dropped here. */
>
>So I would say that whole script does not stop executing,
>only those commands that failed will ... errr... fail.
>
>Should not I expect that not asking not umambuiguous questions is not good
>? :-)

Yes, I definitely wouldn't disagree. <g>

Nico, listen to Ivan. I've put you awry because of my habit of always
running scripts through a TIB_Script which (as you'd know, being an IBO
user) has exception handling built in.

btw, since you do have IBO, why not write a small exe for your customer to
run, using a TIB_Script, instead of making him do it through isql?

/heLen