Re: Validator functions implementation

[Andrew Janke]

On 1/18/20 1:40 AM, Kai Torben Ohlhus wrote:
> On 1/17/20 1:23 PM, Andrew Janke wrote:
>> Hi Octave folks,
>>
>> Are you interested in having an Octave implementation of the Matlab
>> "validator" functions - mustBeNumeric(), mustBeNonempty(), and friends?
>> I have a Matlab-based implementation of them that wouldn't be hard to
>> port over, maybe in time for Octave 6.
>>
>> https://github.com/apjanke/matlab-validoozy
>>
>> Let me know if you're interested, and if so, whether you're interested
>> in my extensions (the "label" argument and the additional validators) or
>> just base Matlab-compatible implementations.
>>
>> Cheers,
>> Andrew
>>
> 
> Hi Andrew,
> 
> Nice to read about another nice "*oozy"-project =)  At least the Matlab
> compatible set of functions was interesting.  With non-compatible
> extensions Octave had some bad experience in the past and should only be
> added with some more care taking beforehand. 

Okay. I'll whip up a subset that just matches the Matlab functionality.

> I noticed that those
> functions are part of a "new" (2017a) classdef syntax [1] but can also
> called individually.

Yep. I've found them very useful for direct calling even without the new
classdef syntax; that's primarily how I use them in Matlab, actually.
About half my function definitions start with "mustBe*" calls these days.

> If you have lots of spare time, please provide a patch with those
> functions using Octave m-style [2] in our tracker.

Done. https://savannah.gnu.org/bugs/index.php?57627

Cheers,
Andrew