Hi Thomas,
> A patch with simple test cases would be nice (maybe a simple SQL script
> with a few cases). I would like to test this with other databases as well.
>
A simple test case where all columns should be TRUE:
SELECT
(CAST('-0.0' AS DOUBLE)) = (CAST('0.0' AS DOUBLE)),
((CAST('-0.0' AS DOUBLE)) < (CAST('0.0' AS DOUBLE))) = 0,
((CAST('-0.0' AS DOUBLE)) > (CAST('0.0' AS DOUBLE))) = 0,
ABS(CAST('-0.0' AS DOUBLE)) = (CAST('0.0' AS DOUBLE)),
ABS(CAST('0.0' AS DOUBLE)) = (CAST('-0.0' AS DOUBLE)),
ABS(CAST('-0.0' AS DOUBLE)) = (CAST('-0.0' AS DOUBLE)),
CAST('-0.0' AS DOUBLE) = (CAST(-1 AS DOUBLE) * CAST(0 AS DOUBLE)),
(CAST('0.0' AS DOUBLE) = (CAST(-1 AS DOUBLE) * CAST(0 AS DOUBLE))),
((CAST(-1 AS DOUBLE) * CAST(0 AS DOUBLE)) < CAST('0.0' AS DOUBLE)) = 0
Of course, same test works with REAL type.
About the patch, I am not sure where I should look. I guess the idea is to
prevent a negative zero to be produced wherever a DOUBLE/REAL is created.
Any hints?
> The question is, should this change be done in a minor version? I think yes
>
Me too. It is causing various unexpected problems, even more so when
combined with the ABS() sign-changing bug, and should be fixed in a stable
version.
As a side note, I mentioned that the point of signed zero was to produce
+inf and -inf when dividing 1 by zero, which Java does. H2, on the other
hand, raises an error about division by zero. So I guess we are on the
right track in removing the signed zero notion :)
Cheers,
-Christopher
--
You received this message because you are subscribed to the Google Groups "H2
Database" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to h2-database+unsubscr...@googlegroups.com.
To post to this group, send email to h2-database@googlegroups.com.
Visit this group at http://groups.google.com/group/h2-database.
For more options, visit https://groups.google.com/d/optout.
