displaying structure arrays

[Thorsten Meyer]

Francesco Potortì wrote:
>> | It seems to me that displaying structure arrays has changed in 3.1.
>> | Now, I get this:
>> | 
>> | octave> x = 1:10;
>> | octave> N = length(x);
>> | 
>> | octave> y = repmat(struct('field1',[],'field2',[]),1,N)
>> | y =
>> | {
>> |   1x10 struct array containing the fields:
>> | 
>> |     field1
>> |     field2
>> | }
>> | 
>> | I think that this behaviour is due to this change:
>> | 
>> | 2008-03-07  John W. Eaton  <jwe at octave.org>
>> | 
>> | 	* ov-struct.cc (octave_struct::print_raw): Don't print contents fo
>> | 	struct arrays that have more than one element.
>> | 
>> | 
>> | How can I see the fields' contents all at once?
>>
>> If you think the behavior in the development version is bad, then I
>> think we should discuss what it should do on the maintainers list.
> 
> In fact, I think it is better than before, as a default.  However, there
> are times I would like to have a quick look at the whole contents of the
> struct array (for small ones, at least).
> 
> Maybe disp() could behave differently and show everything?  Or there
> could be a function controlling the printout, in the spirit of
> struct_levels_to_print() ?
> 
Currently, I am updating the documentation for container types. Due to the above
change in displaying structure arrays, many of the examples for structure arrays
in doc/interpreter/container.txi can no longer be reproduced with the current
development version of octave. And, indeed, I find it quite difficult to produce
good examples to explain structure arrays without a way to display the content
of a complete structure.

Overall, we have this display behaviour for the container types in octave:
 - structures can be printed up to a certain level controlled by the
struct_levels_to_print function
 - cell arrays are always shown fully (as far as I know)
 - structure arrays are only shown to the first level (i.e. field names) now

Actually, I would vote for a consistent way to control the display depth in all
three cases. So what do you think about this:
We replace the struct_levels_to_print function (and internal variable) by a
levels_to_print function that controls the printing depth in all three cases?

regards

Thorsten