Re: Tests for slicing of sparse matrices

[David Bateman]

Kim Hansen wrote:
> On Jan 8, 2008 9:31 AM, David Bateman <address@hidden> wrote:
>   
>> Kim Hansen wrote:
>>     
>>> I wrote some test for slicing sparse matrices in December, here they are.
>>>
>>> They have 9 expected errors where sparse and full matrices do not do
>>> the same. Should I rewrite those tests in order to not call it
>>> expected errors, if that is the behaviour we want from the sparse
>>> matrices?
>>>
>>>       
>> The reason for the 9 failing tests is that this is the behavior of
>> Matlab itself and so they can't really be said to be failing. Therefore,
>> I'd say comment out the failing tests before this is included in octave
>> itself. Also all of these tests are real matrix only and it probably
>> makes sense to have at least complex versions of the same tests. Though
>> the code that these tests exercise is all in Sparse<T>::assign and
>> Sparse<T>::assign1 functions so the same code is exercised in both case,
>> this might not always be the case, and having the same tests with
>> different matrix types might avoid issues in the future.
>>     
>
> The 9 xtests has been changed to asserts that tests positive against
> the current behaviour of octave.
>
>
> 2008-01-10  Kim Hansen  <address@hidden>
>
>       * test/test_slice_sparse.m: Tests for slicing of sparse matrices
>   
Except that perhaps we should have at least complex versions of these
tests, this is a valuable addition and should be included in the test
suite. John can we commit this file?

D.

-- 
David Bateman                                address@hidden
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 6 72 01 06 33 (Mob) 
91193 Gif-Sur-Yvette FRANCE                  +33 1 69 35 77 01 (Fax) 

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary