At 08:16 PM 23/04/2003 -0300, you wrote:

>Hi,
>
>At April 23, 2003, 19:33, Helen Borrie wrote:
>
> > Daniel,
> > Do you have time to test your fix on a column name like "FIRST
> NAME"? Such
> > an identifier is valid inside dblquotes. I suspect that the problem will
> > persist in a case like this...
>
>Actually I've sent another possible fix to Jason. And, "FIRST NAME"
>would be recognized as a field name, see below for further information.
>
>The modifications are in the function GetSQLFirstAndSkip and its local
>function getValue. The changes checks if the return value from
>getValue starts with '(:', ':', '(' or a number. Those are the
>possibilities that I was able to think of. This change works for me,
>but I know that you'll want to test it further to make sure it works
>100% of the time.
>
>After reviewing, there still are some cases where this code might not
>work and would have to be taken into consideration (i.e.: FIRST10 and
>"FIRST 10" as field names). Those are the only 2 type of samples that
>I can think of, that wouldn't work. I you want to I can send you
>privately the IB_Parse.pas that I sent to Jason. It's just what I
>noticed is that it would transform FIRST10 into FIRST 10, unless it
>would be mandatory that there is a space between FIRST and 10. The
>validation on the return value of getValue that would be added is as
>follows: if the first character of the return value is numeric, then
>check to see if the previous character is a space, if it is not a space
>then it is a field name and not the predicates FIRST or SKIP. But,
>still after this additional check, there's still the scenario for the
>field name "FIRST 10", unless it would be to verify that the character
>preceeding the word FIRST is a double quote, and if it is so, then it
>is a field name and not the predicate FIRST. And, what about someone
>using FIRST as a parameter(i.e.: ':FIRST'). I would be able to make
>these modifications tomorrow(it's 8pm for me right now).
>
>Other than these modifications, I wouldn't know what other scenario
>would have to be considered. Am I missing some.

Daniel,
Great work! I haven't had time yet to dive into ib_parse in depth so far
in relation to this problem so, if not too much trouble for you to cc to
me, I'd really appreciate looking at your code changes so far.

regards,
Helen