bug in conv() [changeset for deconv()]

[Ben Abbott]

On Sep 29, 2008, at 9:24 AM, Sergei Steshenko wrote:
>
> --- On Sun, 9/28/08, Ben Abbott <bpabbott at mac.com> wrote:
>
>> From: Ben Abbott <bpabbott at mac.com>
>> Subject: Re: bug in conv()
>> To: "dph.work" <dph.work at gmail.com>
>> Cc: "Octave list" <bug at octave.org>
>> Date: Sunday, September 28, 2008, 5:59 PM
>> On Sep 28, 2008, at 4:47 PM, dph.work wrote:
>>
> [snip]
>> Question for the list; As 3.0.3 is soon to be released,
>> should that be
>> fixed?
>>
>> Ben
>>
>
> Why shouldn't it be fixed ?
>
> Is the fix likely to break other things ?
>
> Thanks,
>  Sergei.

Good question. I hadn't considered that. Thanks!

With the patch in place I ran "make check"

   [...]
   scripts/polynomial/deconv.m ............................ PASS     
4/5    FAIL 1
   [...]

 From fntests.log

    ***** test
   [b, r] = deconv ([3, 6], [1; 2; 3]);
   assert(b == 0 && all (all (r == [0, 3, 6])));
   !!!!! test failed
   operator -: nonconformant arguments (op1 is 1x3, op2 is 3x1)

The row/column orientation of the output of Matlab's deconv() respects  
the first input. I also noticed that Matlab's remainder is always the  
same length as the first input.

Changeset attached.

Ben

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: changeset-deconv.txt
Url: https://www-old.cae.wisc.edu/pipermail/bug-octave/attachments/20080929/0dd99c7c/attachment-0001.txt 
-------------- next part --------------