Turning off auto-grouping of lines into lineseries

Hi! I've painfully stumbled across the following very baffling and infuriating "feature": it would appear that upon adding a fifth line object to a single axis, MATLAB unilaterally and without user notification groups up all existing and subsequent lines on that axis into a lineseries, which then prevents the brush on command from providing those lines with the BrushData property! To whit: I have an axis w/ two lines on which brushing and brush linking was working fine. I added a bunch more lines and suddenly a BrushEnd event handler I have returns the error "??? Error using ==> set There is no 'BrushData' property in the 'lineseries' class." Several hours and much hair wringing later, I finally try adding just one more line at a time: add one more line and it works fine. Add another line, works fine. Add one more line and "??? Error using ==> set There is no 'BrushData' property in the 'lineseries' class." CAN IT REALLY BE THAT AFTER ONLY FOUR LINES MATLAB UNILATERALLY AND WITHOUT USER NOTIFICATION GROUPS UP LINES INTO A LINESERIES OBJECT, WHICH THEN PREVENTS BRUSH ON FROM BEING ABLE TO PROVIDE THEM WITH THE BRUSHDATA PROPERTY!!?? AAARRGGGHH!!?? HOW DO I TURN THIS CRAZY THING OFF (or at least raise the limit at which it operates)?

Answer 1

[Walter Roberson]

Lines are not autogrouped into lineseries.

I happened to find your discussion in a different thread and answered there first. See here

[David Goldsmith]

Can you explain the phenomenon I'm observing, namely that up to a certain number of lineseries (I'm always using plot) added to an axis, brush on adds the BrushData property to them, but above that number, it doesn't?

[Walter Roberson]

I've never used BrushData, and at the moment I don't have a (useable) graphic connection to my server that I could experiment with, sorry.

Answer 2

[Matt Fig]

I am a little confused. All of the objects returned by the PLOT function are lineseries objects. Can you give an example FOR loop which shows the error? Something like:

for ii = 1:5
   L(ii) = plot(1:ii);
   set(L(ii),....)
end

[David Goldsmith]

I'll try (I have a few more things I've thought of to try; if they don't work, I'll try to generate a "minimal reproducer.")

Answer 3

[David Goldsmith]

It appears to be an undocumented (or at least I can't find it documented) feature of brushing, to whit:

 x=-1:0.1:1;
 y=x'*x;
 figure
 plot(x,y)
 brush on
 brushlines=findobj(gcf,'-property','BrushData');
 length(brushlines)

ans =

    20
 hold
Current plot held
 brush off
 brushlines=findobj(gcf,'-property','BrushData');
 length(brushlines)

ans =

    20
% ??????????????????
 plot(y,x)
 brush on
 brushlines=findobj(gcf,'-property','BrushData');
 length(brushlines)

ans =

    20
 lines=findobj(gcf,'Type','line');
 length(lines)

ans =

    42
% Maybe it depends on when one turns brushing on?
% Let's plot(x,y) and plot (y,x) then turn brushing on:
 clf
 % make sure I haven't inadvertently corrupted my data
 size(x), size(y)

ans =

     1    21

ans =

    21    21

[David Goldsmith]

Sorry, my browser's acting up on me:

plot(x,y)

hold

Current plot held

plot(y,x)

brush on

lines=findobj(gcf,'Type','line');

length(lines)

ans =

42

brushlines=findobj(gcf,'-property','BrushData');

length(brushlines)

ans =

0

% !!!!! Even I wasn't expecting that: before when I did it I got 20

Clearly I don't understand this brushing w/ multiple lines very well, but just as clearly, IMO, its behavior is very unintuitive!

In any event, I solved my problem at hand: the extra functions I needed to plot were constant functions, so naively, I was doing plot(vector, scalar), which was generating a new line object for each point! I switched to plot(vector, constant-valued vector) so each new function generated only one line object and this kept the total down sufficiently that brush on is working again. Nevertheless, I'm quite (re-)disillusioned...