Re: problem plotting "N-d object"

[Ben Abbott]

On Feb 17, 2012, at 6:19 PM, Ben Abbott wrote:

> 
> On Feb 17, 2012, at 12:04 PM, Ben Abbott wrote:
> 
>> 
>> On Feb 17, 2012, at 11:59 AM, Jordi Gutiérrez Hermoso wrote:
>> 
>>> On 17 February 2012 11:24, Sergei Steshenko <address@hidden> wrote:
>>>> IMO in Octave/Matlab world the difference between line and column
>>>> vectors is a nuisance in many many cases, and I mention vectors
>>>> because of "transpose not defined" (specifically, "transpose") in
>>>> the original error message.
>>> 
>>> Numpy does have true 1-dimensional vectors, but it still thinks that a
>>> an array of dimensions [1,1,3] is 3-dimensional. You might be happier
>>> with Numpy than Octave.
>>> 
>>>> I think Octave should automatically apply '(:)' operation whenever
>>>> row/column trait doesn't matter.
>>> 
>>> It is difficult to predict in advance when does it matter if a vector
>>> is row or column, so we can't do this everywhere. We do do it in
>>> several instances, but each case requires individual consideration.
>>> 
>>> I don't think "isvector" should return "true" if there is only one
>>> non-singleton dimension. Perhaps it should, but this would break
>>> Matlab compatibility. You might want to make more liberal use of the
>>> "squeeze" function if leading singleton dimensions are causing you
>>> frequent grief.
>>> 
>>> - Jordi G. H.
>> 
>> Jordi,
>> 
>> Any concerns about something like the attached ?
>> 
>> Ben
> 
> The functions which would be impacted are 
> 
>       loglog.m
>       plot.m
>       plot3.m
>       polar.m
>       semilogx.m
>       semilogy.m
> 
> Would squeezing any of the inputs whos ndim > 2 break anything ?
> 
> I don't think so. As I have time, I'll modify the texinfo and add some demos.
> 
> Ben

I've added a feature request and attached a changeset.

        https://savannah.gnu.org/bugs/?35566

Ben