From: "Sung Soo Kim" <>
Newsgroups: comp.soft-sys.matlab
Subject: Re: Very weird resolution issue. Bug ????? Seriously !!
Date: Wed, 4 Mar 2009 03:25:03 +0000 (UTC)
Organization: JHU
Lines: 12
Message-ID: <goksaf$nfd$>
References: <goju5b$nv4$> <gok3qm$2gn$> <gok5db$mqi$> <gok6c5$19g$> <gok8gq$2na$> <gok9cs$3ut$> <gokdr9$heh$> <gokgtb$hfg$>
Reply-To: "Sung Soo Kim" <>
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: 1236137103 24045 (4 Mar 2009 03:25:03 GMT)
NNTP-Posting-Date: Wed, 4 Mar 2009 03:25:03 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1724905
Xref: comp.soft-sys.matlab:522322

> The way I look at it, we have two choices when dealing with floating point number equality computations:
> 1.    Rely on them and be surprised when they don't work.
> 2.    Don't rely on them, work around it, move on.
> I still think calling this a 'bug' is inaccurate, but really (and again) this is something you should bring up with TMW directly.

Well, I didn't rely on it. My test cases are composed of tests that uses '==' or tolerance. All of them were fine. I decided to use '==' when it actually showed equal result. But it suddenly changed in later versions. It may be related to MATLAB's code optimization algorithm. It is annoying to see the change that was not a bug at all and was something right.

But, Matt, I think your view is very balanced, and I admit that I'd better follow your advice. I'll report it but I'll not rely on that.

Thanks a lot.