Wed 31 Dec 2014 07:21:59 PM UTC, comment #6:
Fixed on the development branch in this changeset (http://hg.savannah.gnu.org/hgweb/octave/rev/e5a78897be9e). Closing report.
|
Tue 30 Dec 2014 12:07:42 AM UTC, comment #5:
Matlab says:
>> x = {''}
x =
{''}
|
Mon 29 Dec 2014 10:57:33 PM UTC, comment #4:
Another Matlab test:
Does this produce an empty cell or a 1x1 cell with an empty entry?
|
Mon 29 Dec 2014 07:52:48 PM UTC, comment #3:
Oh boy, Matlab compatibility. Can I get a few more examples run? Maybe easiest to invoke 'diary on' before executing all these and then just upload the file rather than trying to use verbatim notation.
|
Mon 29 Dec 2014 05:09:11 PM UTC, comment #2:
MY Matlab R2009b says:
>>z = cell(5,4,3,0,2);
>>{1,2,3; z{:}; 4,5,6 }
ans +
[]] [2] [3]
[4] [5] [6]
>>
|
Mon 29 Dec 2014 04:41:04 PM UTC, comment #1:
Confirmed. I think the general summary rule is that Matlab will allow the concatenation of any empty value because it simply removes it.
Could you try a slight variation of your original test where the empty cs-list is in the middle of the construction?
|
Sun 31 Aug 2014 12:40:59 PM UTC, original submission:
In MATLAB, the following is valid:
As you can see, the cell array whose elements are unpacked doesn't even have to have a size consistent with the target cell, such as 0-by-3 in the above example. This makes sense, as that information presumably isn't present in a cs-list any more.
In contrast, here's the current Octave behavior:
|