[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Gnash-commit] /srv/bzr/gnash/trunk r10536: return NaN from Math.pow
From: |
Sandro Santilli |
Subject: |
Re: [Gnash-commit] /srv/bzr/gnash/trunk r10536: return NaN from Math.pow if first arg is not finite: fixes math-function tests from swfdec |
Date: |
Sat, 10 Jan 2009 10:15:41 +0100 |
On Sat, Jan 10, 2009 at 09:52:32AM +0100, Benjamin Wolsey wrote:
> > return NaN from Math.pow if first arg is not finite: fixes math-function
> > tests from swfdec
> > modified:
> > libcore/asobj/Math_as.cpp
> > testsuite/swfdec/PASSING
>
> Which player does this? LNX 9.0.115.0 and a LNX version 10 player return
> Infinity, not NaN for Math.pow(Infinity, 0.4), which was the reason I
> didn't change it (as I said in the commit note) when fixing all the
> other math tests. The latest Windows player is still the reference,
> isn't it? Can anyone verify that the test is correct in Windows? I think
> we need tests in actionscript.all and a note if there is a difference.
Oops, I can confirm with both LNX versions.
Was blindly trusting the swfdec testcase.
> While we're on this topic, there's a bug open about different matrix
> values in Windows / Linux players (http://savannah.gnu.org/bugs/?24135),
> and that could really be closed too, if we all agree that the Windows
> player is the reference.
I agree on that, the linux version seems often broken.
--strk;
GIS & Flash consultant/developer () ASCII Ribbon Campaign
http://foo.keybit.net/~strk/services.html /\ Keep it simple!