A line can be determined by three numbers

{Number|function} a {Number|function} b {Number|function} c

where a point (x, y, z) with ax+by+cz = 0 lies on that line. But the point (-10, 3, 1) defined in jsxgraph by

board.create('point', [1, -10, 3])

doesn't lie on a plane defined by

board.create('line', [1, 2, 4])

but -10*1+3*2+4*1=0.

If JSXGraph is used together with goole translate the mouse position is slightly off.
Example (replace gaId):

<div id="google_translate_element" class="sup7"></div><script>
function googleTranslateElementInit() {
  new google.translate.TranslateElement({
    layout: google.translate.TranslateElement.InlineLayout.SIMPLE,
    pageLanguage: 'ru',
    gaTrack: true,
    gaId: 'UA-?????'
  }, 'google_translate_element');
}
</script>
<script src="//translate.google.com/translate_a/element.js?cb=googleTranslateElementInit"></script>

A similar issue seems to be known:
https://groups.google.com/forum/?hl=de&fromgroups=#!searchin/google-translate-general/clientX/google-translate-general/H2zj0TNjjpY/jw6irtPlCw8J

We have a division by zero problem in the parabola element

https://github.com/jsxgraph/jsxgraph/blob/fe457c97/src/element/conic.js#L601

If F1 and M are too close, e is too close to zero and the matVecMult() or a following operation causes a crash in Chrome and Opera (apparently not Firefox).

allow the user to load a construction from file synchronously and/or allow the user to specify a callback function

(this is a ticket from a backup of our old tracker at sourceforge trac)

In fact, most of the time it is not working:

http://jsfiddle.net/migerh/3pSVZ/

Pan the view (with shift+drag), especially to the left. At some point the TEST-X1 label will jump up to the correct position.

When a ticks are set onto a line parallel to the y axis the last do-while loop in the updateTickCoordinates method gets stuck in an endless loop, because _tickEndings() always returns an array containing two elements.

Enable hashes to mark parallel or congruent segments.

Gliders on polygons are broken. They are jumpy and can't be set on every spot on the polygon.

For IE9 running in IE7 mode (in an embedded WebBrowser component in a WPF-application) The document.body doesn't seem to be necessarily initialized when the following function is called, resulting in an error saying that : "unable to get value of the property 'scrollLeft' object is null or undefined".

function MouseMove() {
    document.body.scrollLeft;
    document.body.scrollTop;
}

(The little bit of magic in jsxgraph.js, lines 95-98 :) )

This can be avoided just by adding a simple "if(document.body){ ... }" check.

This rather exotic case (IE9 in IE7 mode within a WPF WebBrowser-control, !DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN") is currently the only tested one.

JXG.Options.text.display = 'internal';
If you select a point, its color will turn into red.
It is not a problem in JXG.Options.renderer = 'canvas';

let gliders snap to every intersection point of their glide object and the grid lines

'Additional' properties can be added to the label object itself. There's no need for a subobject named content.

(this is a ticket from a backup of our old tracker at sourceforge trac)

implement hasInnerPoints for conic sections.

See this posting https://groups.google.com/forum/?fromgroups#!topic/jsxgraph/dZ_Bbm1oYpo\

While dragging lines and circles the defining points of the dragged element should snap to grid (if set).

Implement dragging of polygons

new style options for angles:

general: 'sector', 'none', 'square'
right angle: 'sector', 'none', 'sectordor', 'square'

Implement some faces for ticks like <, >, [], ^, v, <>, +, x, and (if possible) o.

It looks like IE10 doesn't support the Touch Event API and instead Microsoft designed and implemented a new Pointer Event API which will be used as a starting point for a new touch event api by the W3C.

Example: http://jsfiddle.net/PNLxv/

If the document body has padding-left and/or padding-top, this gets added to the offset in JXG.Board.getCoordsTopLeftCorner, so the positions of mouse events are incorrect.

I can't see why the function getProp in getCoordsTopLeftCorner uses a variable pCont in the closure to refer to the element to get style from, instead of taking it as an argument. Changing that should fix it, but I can't build JSXGraph on windows, so I can't do it myself.

I have been trying to figure out but all in vain as to why doesn't the features of JSXGraph such as glider and other transformation features do not work with Easeljs. I guess the redrawing of canvas or something is stopping it.

A simple code such as

var p1 = board.create('point',[a/2,-b/2], {name:'A', size:1});
var t = board.create('transform', [Math.PI/3,'A'], {type:'rotate'});
var p2 = board.create('point',[-a/2,-b/2], {name:'B', size:1});

also doesn't transform the two points and draws as if there was no transform statement.

Let the user restrict the grid to a finite area. The restriction should be given as attributes.

Hi,

I have drawn a circle example in IFrame. Its drawn. but enlarging of circle with point on circumference doesn't work.

please reply asap.

thanks

Currently the 'shear' transformation doesn't actually perform any shearing as it is applied to the wrong element in the matrix, instead it results in a scaling operation. Additionally it can only be applied along the X axis.

We don't need preventDefault() and return false in every case, sometimes it would be better to return true and to not prevent default browser actions, e.g. on touch devices when the user wants to scroll the page and touches the board but no objects or if there's a button on the board the user can click.

As sometimes this behaviour might not be desired, it should be configurable via board options.

make the shift+drag panning optional and implement shift+wheel for zooming

With selector methods one could select a bunch of elements and batch-process them, e.g. attach event handlers, update attributes, etc.

If the attracted element is not snapped onto the attractor and the attractor is removed, the attracted element will also be removed.

See

https://groups.google.com/forum/#!topic/jsxgraph/jt0MQrHqsu4
http://jsfiddle.net/BUj86/27/

board.create("segment", [a,b],{dash: 2}); does not work when JXG.Options.renderer = 'canvas';

We have a lot of functions expressions that are used as methods of elements and are defined on the fly in creator functions. With the AMD pattern we can define these methods as local variables of each module and simply reuse the definitions.

We'd have to refactor these functions and replace every access to parameters of the creator functions (primarily the parameters parents and attributes) with properties of the elements.

But e.g. JXG.GeometryElement.addRotation() assumes these methods exist.

Add mouse/touch versions of every element event.

See this posting.

In the GUI we want to place points on all possible kinds of intersections. This works pretty good for lines and circles but intersections with curves are missing.

So we need a function wich computes all (> 2) intersections of a given curve with another object [ like a curve, line or circle ]. These intersections should be CAS points and ideally being returned in an array.

(this is a ticket from a backup of our old tracker at sourceforge trac) - original reporter: @haaner

Structuring the code using the AMD (Asynchronous Module Definition) pattern, and using RequireJS, would introduce the following advantages:

• File dependencies will become very declarative, each file will specify its dependencies, which would make it easier to maintain
• Provides a standardised approach to loading scripts, which would allow third parties to easily integrate it into their own source code
• By using RequireJS, the build process can be simplified by using its built tool that will combine files taking dependencies into account, and apply optimisations using UglifyJS or Google's Closure compiler.
• The build process can be configured to generate a single file and instead of using full RequireJS library, use AlmondJS, which is a "shim" loader.

I am currently considering Forking the project and doing it myself, but to set it properly, it would help to have the tree of dependencies between all files.

It would be nice to place a glider not only on a one-dimensional object, but also on a object of 2 dimensions (OBJECT_CLASS_AREA), so it can be placed in triangles, polygons or circles ...

(this is a ticket from a backup of our old tracker at sourceforge trac) - original reporter: @haaner

I am unable to drag or perform any mouse event on the gliders when my entire object is a child of easeljs exportRoot. How do I resolve this issue?

Thanks
Bharathy

For sigil: Delete all contents of the container div before parsing the code.

For epub: Add a reload button to the navigation bar.

Provide a new event type 'click' and/or 'tap' which is fired when the user clicks/taps on the an element. Possible approach:

Use the existing event handlers and use timing and finger/mouse position information to determine a click situation.

Another, at first glance easier approach would be to register the click event handler on the board container and use existing JSXGraph methods to determine the elements below the mouse and fire a click handler. But this will only work for mouse events because browsers do not offer a tap or 'touchclick' event, only touchstart and touchend.

If a user clicks the navigation controls the events seem to bubble up and cause a mouse down/up on the board, too.

See http://jsfiddle.net/Cleonis/LUnVA/

When a Circle is created, supplying the coordinates with anonymous functions, JSXGraph must create the points in order to create the Circle. Only the circle is shown, of course.

As per my enhancement request (#24) the element 'Sector' now accepts anonymous functions. But the created points are shown.

Enhancement request: make 'Sector' behave like 'Circle': do not show the created points.

I copied the library from Github. When I opened examples\vector.html I can zoom chart with the middle mouse button. This example uses the version 0.92/jsxgraphcore.js
In all other exaples, I can not zoom the charts. And if I change jsxgraphcore.js in vector.html on version 0.97 zoom stops working. How fix it?

Attributes should not be overruled silently. Right now this affects the 'visible' attribute. An element should be visible only if the user says so (attribute 'visible') and there is a representation in the real plane.

If I try to decode characters outside the Basic Multilingual Plane, the decoding function fails.

For example:
JXG.Util.utf8Decode('\xf0\x9f\x99\x8a')

This should be one unicode character, but it decodes to two.

There could be many ways to fix this, but what we have done, is to work around this with this very interesting solution:
http://ecmanaut.blogspot.ca/2006/07/encoding-decoding-utf8-in-javascript.html

so: decodeURIComponent(escape('\xf0\x9f\x99\x8a'))

It is a hack with a deprecated function, but it works for unicode characters in the astral plane and also as the functions are implemented on native code, it is very fast.

see https://groups.google.com/d/topic/jsxgraph/7NUrt9jhTBM/discussion

when having multiple polygons with labels, removing the last one removes the label of the first instead.

The problem is caused because all polygon labels have the same id.
Manually setting an id solves the problem

Allow angles to be displayed clockwise.

The update of the points in a group doesn't work anymore

See:
http://jsfiddle.net/Cleonis/LUnVA/

I assume the intention is that the element 'sector' accepts anonymous functions, just as for instance the element 'circle' accepts anonymous functions.

In the jsfiddle, when the code for 'sector2' is uncommented, jsxgraph reports an error, and the sector is not drawn.

Make the generateLabelValue parameter an attribute so it can be used in initBoard()

The construction of angular bisectors fails in case of parallel lines and snapToGrid enables.

The implementation of element.parents is not clean. For example parents given by name (as string) are ignored.

check setBoundingBox if keepaspectratio takes care of the viewport being centered or not

(this is a ticket from a backup of our old tracker at sourceforge trac)