plot templates and options lists for set, plot etc.

[Ben Abbott]

On Jun 7, 2009, at 4:08 AM, Thorsten Meyer wrote:

> Hi,
>
> Ben Abbott wrote:
>>> Yes, I have seen, that the options are collected into a cell array
>>> within the plot functions. However, keeping the options in a  
>>> structure
>>> has the advantage, that I can easily access and set the individual
>>> options, e.g. to transform one plot template into a new one,  etc.
>>> Also, how are graphic objects realized internally? Aren't they also
>>> structures? It always seemed quite strange to me, that the get()
>>> function returns a structure, while all the functions setting handle
>>> options take lists of key, value,... as input arguments...
>>>
>>> Thorsten
>>
>> Hmmm ... when I first began using set/get, I had assumed that  
>> s=get(h)
>> would be reciprocated by set(h,s). Of course there are problems with
>> that example due to some properties being read-only.
> I have been thinking about implementing a way to save a plot by  
> getting
> all the objects in in with get(h) and saving the corresponding
> structures - and then being able to reproduce it with a sequence of  
> set
> commands. Will that be possible at all considering those read-only
> properties?
>>
>> However, isn't the functionality below what you're looking for?
>>
>> style = struct("linewidth", 2, "marker", "x", "markersize", 12);
>> h = plot (1:10)
>> set (h, style)
> exactly.
>>
>> I don't see any problem with compatibility. If set() were an m-file,
>> such a change would take me a few minutes, but with c/c++, I won't be
>> much help proposing a patch.
> I would like to try to do this, if nobody objects to such a change.
>
> Also a question regarding the implementation:
> I see that in graphics.cc(DEFUN set) the handle is extracted and the
> rest of the arguments are passed to the set method of the graphics
> object. If I want to make set accept also structures, I could either
> handle this in the DEFUN set (expanding the structure to a list of  
> key,
> value and pass the adapted list of arguments to the set method) or
> change the set method itself. Which of the two would be preferable?
>
> Thorsten

I'm moving this discussion to the developers list, and have cc'd John  
and Jaroslav. Perhaps one of them can comment?

Ben