I'm trying to create a plugin for flot that will draw axis that cross the origin (0,0) instead of only being shown at the edge of the plot. One thing I'd like to do is, when using my plugin, the drawing of the regular axis (the ones at the edge) should be suppressed. So I added some logic to a function that hooks into processOptions like this:
plot.hooks.processOptions.push(processOptions);
function processOptions(plot, options) {
if (options.crossOrigin) {
if (options.xaxis.crossOrigin) {
options.xaxis.show = false;
plot.hooks.drawBackground.push(drawXAxis);
}
if (options.yaxis.crossOrigin) {
options.yaxis.show = false;
plot.hooks.drawBackground.push(drawYAxis);
}
}
}
By setting options.xaxis.show to false (and the same for the yaxis), this is supposed to stop the axis from appearing, but it doesn't work. The option gets set, but flot ignores it when it draws the graph. The only way I can get it to not plot the regular axes is to explicitly set the show option when setting up the plot.
Any ideas why this doesn't work? Is there a way to make it work?
Update: So after playing around with this for a while, I think what is happening is that there is, for example, an options.xaxis and an options.xaxes (an array) which contains just 1 entry. For some reason, flot wants to use the 1 axis in the options.xaxes[0] instead of options.xaxis. But I'm not quite sure what the logic here is supposed to be. They both get my custom "crossOrigin" option merged in, but only the options.xaxis gets it's "show" property set.
Using xaxes[0] instead of xaxis seems to fix the problem, although it's not clear when and how flot is processing this stuff.
Related
VB2010 using MS Chart: I've added one series as a point chart. On occasion there are over 1000 points and the point labels become way too jumbled. I have smart labeling ON:
cht.Series("srs").SmartLabelStyle.Enabled = True
But nonetheless it still looks bad. So I added a context menu to turn off the labels. The user can then zoom in to a spot and if they wish turn labels back on. I cant seem to figure out a way to do this without looping through ALL data points.
I can totally hide the points and labels by
cht.Series("srs").Enabled = False
But I want just the labels to be hidden and then redisplayed when the user has opted for it.
Any help appreciated.
Edit:
Since I haven't found a method to turn labels off and on with one command I am currently cycling through each point in the series.
Me.Cursor = Cursors.WaitCursor
Application.DoEvents()
'suspend updating UI
cht.Series.SuspendUpdates()
'cycle through all points in the series and set the label either to an empty string or whatever is cached in the Tag property.
'todo: this is not efficient for large datasets but its the only thing we have.
For Each pt As DataPoint In cht.Series("srs").Points
If mnuDisplayLabels.Checked Then
pt.Label = pt.Tag.ToString
Else
pt.Label = ""
End If
Next pt
'resume updating UI
cht.Series.ResumeUpdates()
'force redraw of chart
cht.Update()
I think you have to loop, but you want to suspend updating the UI until you've done it for all points. Try something like:
chart1.Series.SuspendUpdates();
foreach (Series s in chart1.Series)
{
s.IsValueShownAsLabel = false;
}
chart1.Series.ResumeUpdates();
When I create a figure with MATLAB with a title, then use the File|Print option to print the figure, the title is clipped. Please try this code for an example
t = linspace(0,2*pi,1000);
s = sin(t);
figure
plot(t,s)
titleString = sprintf('Multi\nLine\nTitle');
title(titleString)
disp('Now press File|Print Preview on the figure and observe that the title is clipped.')
disp('This happens with all titles, the multi line title makes it more obvious.')
disp('I know I can fix it with Fill Page or Center, but I should not have too.')
You can also see the problem in print preview. As I say in the example code, I know I can get round the problem using Print Preview then Fill Page or Center, but I don't want people using my code to have to use a work around.
I have observed this problem with r2014a and r2015b. I assume other releasse are also affected.
Is there setting I can make before creating the figure that centres the plot or fills the page and makes the problem go away? Is there some other setting I should make to avoid the problem?
Here is a little more debug information. If I press File|Print Preview, MATLAB reports Left 0.64, Top -0.59, Width 20.32, Height 15.24. I guess the problem is related to the negative Top value. These are defaults from MATLAB; I have not made any attempt to change these values.
One extra thing. I am in the UK, so my default paper/printer setting will be for A4 paper, if that makes a difference.
Edit:
It looks like my problems are caused by two lines further up in my program:
set(0,'DefaultFigurePaperOrientation','landscape')
set(0,'DefaultFigurePaperType','A4')
I think that becuase plots expect to be on paper with a portrait orientation, I am seeing these problems.
Perhaps I should revise my question to: what to I need to change in MATLAB figures so they print correctly on landscape A4 paper (ideally in the center, scaled to fill the page, but with correct orientation). All this without using Print Preview.
But I am going to do this instead to code around my problem.
set(0,'DefaultFigurePaperOrientation','portrait')
set(0,'DefaultFigurePaperType','A4')
I can't seem to reproduce your problem on the computer I'm currently on (see the values I'm getting by default - Top is 8.11):
However, if your problem is what I think it is (I'm getting something that fits this description on another computer I'm working on), try adding _{ } at the end of your string. This is a TeX string meaning "subscripted space" which pushes the rest of the text slightly upward. You can also use ^{ } on the first line if the clipping is happening from the top. I found this workaround to work on axis titles and labels as well.
Exaggerated, the workaround looks like this:
titleString = sprintf('^{^{^{^{^{^{ }}}}}}Multi\nLine\nTitle');
Which shows the word "Multi" even for Top = -0.59.
If the above is not what you're looking for, you might want to look at the robust export_fig.
I can confirm that my clipping problems are caused by this line:
set(0,'DefaultFigurePaperOrientation','landscape')
I have revised my program to start with this instead.
set(0,'DefaultFigurePaperOrientation','portrait')
set(0,'DefaultFigurePaperType','A4')
And the problem has gone away.
Users can still print in landscape if they use the print preview feature.
Today i tried adding a slider below my graph, so different graphs would be drawn when you slide it. Eventually i got all of this to work perfectly fine, however i still have two small issues.
On top of the figure with the slider, the figure toolbar is gone. It can easily be added by clicking view -> figure toolbar at the top in the figure itself. However to do this everytime is a bit tedious. And when i google'd for the problem, the results where are all guides to create custom toolbars. I just want to add the default toolbar back, it was fine. But i do not know how to do that. Typing 'uitoolbar' gives an empty one. But i am not sure how to fill it with the default icons.
And the second problem i ran into;
The main script calls in the end for a function, this function then creates the new graph with the slider. However herefore i needed to transfer the matrix of data to plot from the main script to the function, so i made the matrix a global variable. However as soon as the function is called, matlab gives me this error (once for each global variable):
> In xxxx at 270
Warning: The value of local variables may have been changed to match the globals.
Future versions of MATLAB will require that you declare a variable to be global before you use that variable.
Before i call the function i type
global xL yL zL p
which are the data required to plot. Then the first line in the function file is this same line again. The data seems to be transmitted fine and the graph shows the correct plot.
Is this some error i can just ignore then? Since everything works as it should? Or what does it mean exactly?
Ok here are my 2 cents:
1) From what I understand the simplest way to fetch the toolbar would be using findobj and then make it visible:
hToolBar = findobj('Type','uitoolbar');
set(hToolBar,'Visible','on');
You can also get the toolbar's properties using "get". Here is what it looks like from a simple script I ran:
get(hToolBar)
hToolBar =
0.1890 %// don't worry about that number; it's the handle for the toolbar.
BeingDeleted = off
ButtonDownFcn =
Children = [1.18896]
Clipping = on
CreateFcn =
DeleteFcn =
BusyAction = queue
HandleVisibility = on
HitTest = on
Interruptible = on
Parent = [1]
Selected = off
SelectionHighlight = on
Tag =
Type = uitoolbar
UIContextMenu = []
UserData = []
Visible = off
I don't know why it disappears though.
2) You might get rid of those global variables if you sent your data as inputs to your function is that a possibility?
For example:
function PlotData(xL,yL,zL,p)
%// Your code here
Hope that helps!
Hi everyone and thanks in advance for all your help.
I'm currently working on a UI in matlab that includes 4 differents axes to show various info. Basicly, I have MRI data (so 3D data) that I want to show. I use 1 axe to show 3 perpendicular plane that split the data at its center (axeMain), and 3 other axes that shows the planes individually (axeX,axeY and axeZ). I have a main axe on wich I want to make rotations possible (only on the main axis system) and I want to be able to select coordinate on the 3 other axe (those with 1 plane only). I also unable translations but that is out of scope for my problem.
So basicly, I have no problem selecting my coordinates (using the buttondownfcn on my planes) on all of the 3 axes and I also have no problem using rotate3D on the mainAxe. HOWEVER, if I try to have both to work at the same time, the buttondownfcn doesn't work anymore. I don't understand why it's doing this (I have some ideas but that's about it) and I have no idea how to work around it. Basicly my code for those functions are like this :
rotate3d(handles.axisMain);
%some other code, setting up other UI stuff
%allow selection on the 3 static plains.
set([handles.axeX,handles.axeY,handles.axeZ], 'buttondownfcn', #getCoord);
So my question is basicly : why wont those 2 functions don't work together and how to work around that problem, ideally with minimal code change?
Thanks.
EDIT : this is a print screen of my current interface. I want to be able to select coordinate using ONLY on the 3 last axes (containing only 1 plane in each of them). I also want to be able to rotate images ONLY on the first axe (the one with 3 planes). Hope this clarify.
I would also like to note this : following my tests, I found that the mouse click would not raise at all if the rotate3D is activated. The problem is not in the logic I use to get the coordinates itself, but in the click event not being fired.
Also, I am aware that the event is not on the surface that I try to print. The actual code is like this :
set(h, 'buttondownfcn', #getCoord);
set(h,'HitTest','on');
where h is the handle of a surface and each surfaces are processed this way.
I have found the exact solution!
Exploring the code for rotate3d(hAxe,'on') , i have found that they key line is getuimode(hFig,'Exploration.Rotate3d') wich returns uitools.uimode that contains ButtonDownFilter and ModeStateData methods.
Finally i have solved the problem in the following way:
rotate3d(hAxe,'on')
uiMode = getuimode(hFig,'Exploration.Rotate3d');
uiMode.ButtonDownFilter = #mycallback;
where:
hAxe is the handle to the axes object
hFig is the handle to the figure object
#mycallback is the callback that lets buttondownfcn and rotate3d works together as in the example of Matlab's help
So, you have to tag the object that you want to not be ignored:
set(hAxe,'tag','DoNotIgnore')
And write the callback:
function [flag] = mycallback(obj,event_obj)
% If the tag of the object is 'DoNotIgnore', then return true
objTag = obj.Tag;
if strcmpi(objTag,'DoNotIgnore')
flag = true;
else
flag = false;
end
maybe
set(handles.axisMain, 'buttondownfcn', #getCoord);
To try and solve this problem, I started looking into writing a plugin for flot to do what I needed. It looked surprisingly straight-forward, but I have run into a minor snag. I First used the draw hook to draw my axis (note: I'm aware this code needs some error checking to make sure the origin isn't off the chart)
function draw(plot, ctx) {
var offset = plot.getPlotOffset();
var origin = plot.pointOffset({ x: 0, y: 0 });
ctx.moveTo(offset.left, origin.top);
ctx.lineTo(plot.width() + offset.left, origin.top);
ctx.stroke();
ctx.moveTo(origin.left, offset.top);
ctx.lineTo(origin.left, plot.height() + offset.top);
ctx.stroke();
}
And this works, with 2 small snags. The first and most important is that for some reason it makes the very last point of my data series change color! See here:
The point in the upper right-hand corner has turned black. By stepping through the code I can observe that this only happens when the axes are drawn. Before my draw code is called, the point is the correct color. So I think it has something to do with the drawing on the first "stroke". For some reason it seems to then apply the black color to the last point. Any idea on how to fix this?
The second problem is that this code is called only after flot has draw everything else (including the data points) which means that my lines appear on top of points instead of underneath them. This is only a problem for those two points that are very close to the vertical axis. Looking at the flot documentation, it mentions a drawBackground hook, but this doesn't seem to actually exist. I tried using it and it complained the drawBackground was null. So what happened? And is there a better place to hook into and draw my axes? I want to draw after flot has worked out the size of the plot area and before it draws the first series of data points.
What's currently on the website API documentation does not match the newest released version (0.7). In order to use the backgroundDraw hook, you'll need a newer version of flot, directly from their github repository. The newest version is fine, but anything after this commit would be workable.