This #@!* bug take all my happiness because it's very hard to reproduce it ! sometime it's happen in few hours, sometime no probleme for one or 2 weeks... my mistake was to upload a database top clean that just come from a restore, before i have the same database but little used (because of my test on it, but no probleme in validation) and the bug appear in one night in average on this database... stupidely i delete this database to upload only clean data to not gave false positive... i hope vlad will be able to reproduce the bug this week and that he have one free computer to let my demo app working .... and i very very (but very) thanks vlad for his wonderfull support (and you too) !!

anyway speaking about bug i have another more easy to reproduce and critical too :

in FB 2.5 RC2 it's work perfectly but in the last FB 2.5 snapshot i have an error

try this :

CREATE TABLE TEST (
ID VARCHAR(100),
Longitude NUMERIC(11,8),
Latitude NUMERIC(10,8)
);
COMMIT;

INSERT INTO TEST (ID, LATITUDE, LONGITUDE) VALUES ('IT-00298601-00298707-00298781', 41.218463, 14.528778);

commit;

Select
ACOS((SIN(latitude*(0.0174532925)) * SIN(0.71939789)) +
(COS(latitude*(0.0174532925)) * COS(0.71939789) * COS(0.25357501 -
(Longitude*(0.0174532925))))) * 6371
from
TEST


from FB2.0 to FB 2.5 RC2 no problem
in FB2.5 last snapshot in have an error saying that Acos must be between [1 and -1]


stephane

--- In firebird-support@yahoogroups.com, "Ann W. Harrison" <aharrison@...> wrote:
>
> Stephane & Vlad
>
> It makes me very happy and hopeful for Firebird to
> read your messages as you work on reproducing and
> understanding a hard bug. (Yes, Stephane, in
> most circles, a bug that takes two days to reproduce
> counts as hard.) This is the way open source is
> meant to work.
>
>
> Cheers,
>
>
> Ann
>