Up Prev Next
From: Greg Earle <earle at isolar dot DynDNS dot ORG>
Subject: Re: DBD::Sybase 1.07 install on Solaris 8 - failing t/exec.t test #11
Date: Nov 2 2005 10:11PM
On Nov 2, 2005, at 1:42 PM, Steve Baumgarten wrote:
> Greg Earle wrote:
>>> Floats tend to do this from time to time.
>>> Reason I tend use decimal instead.
>> So, what's the consensus - do I just change the test in exec.t
>> to match my returned mutant results? :-) And do a "make test"
>> and "make install" manually, ignoring the mismatched results?
> Mix floats and the == operator and you're bound to be surprised sooner
> later. One must compare to a certain desired precision.
On Nov 2, 2005, at 2:05 PM, Ravikumar Srinivasan wrote:
> Floats don't like you it seems. ;)
Agreed - but the kicker is, I seem to be the first person in the
known (== Google :-) ) world to have stumbled across this. Is it
due to some sort of setting in the Sybase server (which I don't
control, I'm not the DBA) that causes it to return this long
float answer which fails the matching test?
I'll try your test code to see if I can massage exec.t into
playing nice so I can get DBD-Sybase to install - I've got
other fish to fry with this new Perl 5.8.7 setup I'm trying
to add DBD-Sybase and DBD-Oracle to.
(Ravikumar, good suggestion - I'm just nervous that if this
one test fails, then something else might be broken in actual
DBD-Sybase usage. So I'm trying to get to the bottom of this.)