[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Octave-bug-tracker] [bug #41831] image 2.2.1 - miscomilation of bwdist.

From: Carnë Draug
Subject: [Octave-bug-tracker] [bug #41831] image 2.2.1 - miscomilation of bwdist.cc on i386
Date: Tue, 11 Mar 2014 16:20:28 +0000
User-agent: Mozilla/5.0 (X11; Linux i686; rv:24.0) Gecko/20140207 Firefox/24.0 Iceweasel/24.3.0

Update of bug #41831 (project octave):

                  Status:                    None => Fixed                  
             Open/Closed:                    Open => Closed                 


Follow-up Comment #1:

Hi Thomas, thank for you reporting this bug.

I have already fixed this with cset 43cd1c5e0955. As you, I couldn't find any
issue with the code and could only reproduce on i686. I found that changing
those variable to double fixes it. Also, changing the sqrt2_1 inside
quasi_euclidean() to double would also fix it. I still think there must be a
proper fix somewhere else, but for the time being, this seems to work.

Also, this bug happens in one of the tests, but it was luck. When I tried with
random matrices, only some of them would trigger it.

By the way, this problem should already been present on the image package
version 2.2.0. It was caused in 27d87f06dd99 when I changed from double to
single for Matlab compatibility.


Reply to this item at:


  Message sent via/by Savannah

reply via email to

[Prev in Thread] Current Thread [Next in Thread]