Remove data when an element is removed

[krampstudio]

As a user I'd expect that when an element is removed the .data() linked to this element are also removed (or at least flushed when Paper.clear() is called).
Nevertheless, if I create another element with the same id, I'm getting the data of the previous element (due to the way it is stored inside the variable eldata)

May be removeData could be called inside elproto.remove

[tomasAlabes]

Now ids are random, so at least you shouldn't have that problem.

[jvilk]

Note that this bug causes a memory leak, as the eldata variable grows without bound as new data is added without being removed.

[davidjmcclelland]

I am seeing an issue that tracks with what @jvilk is describing, specific to IE9-Edge where memory is not released because objects are retaining references to data. Other browsers do not seem affected. Would love to fix this if it doesn't require a ton of changes.

[jvilk]

@davidjmcclelland I have a fix open in a PR here: #1077

It's a one line fix that's easy to apply.

[davidjmcclelland]

Thanks for contacting me John! I just checked the source in the version I am using and it has this change in it. It has a comment that indicates it is version 2.1.0, but I have found numerous undocumented changes in the code done before I got involved. I am still having memory leaks big time in IE 11/Edge - they don't use VML, couldn't have ever been tested with SVG... MAC http://davidjmcclelland.com | http://twitter.com/davidmcc | LinkedIn: http://www.linkedin.com/in/davidjmcclelland

…

On Thu, Apr 6, 2017 at 12:20 PM, John Vilk ***@***.***> wrote: @davidjmcclelland <https://github.com/davidjmcclelland> I have a fix open in a PR here: #1077 <#1077> — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#870 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AANwnedprnHrvyxGf12K77BenCn1VX-3ks5rtRDDgaJpZM4CAi9z> .

[jvilk]

@davidjmcclelland is this code open source, by any chance? (I'm not promising anything, but I've been on a bit of a JS memory leak bent.)

And is this browser specific? That sounds odd.

[davidjmcclelland]

Not open source. Someone took to hacking the Raphael library source in the project I support at some point a few years ago. With IE 11 / Edge switching from VML to SVG, it isn't too surprising to me that there might be a leak - I think IE does a different kind of GC than other browsers. I can't swap out the polluted library because then our visualization break - ironically if produces leaks in all browsers without the hacks in place! Maybe that's how your PR got in there - there are changes cherry picked from several subsequent releases in our special version : ) Its very frustrating because the Dev tools in IE11 are still not very good and the Raphael library is not easy for me to grok. MAC http://davidjmcclelland.com | http://twitter.com/davidmcc | LinkedIn: http://www.linkedin.com/in/davidjmcclelland

…

On Thu, Apr 6, 2017 at 9:54 PM, John Vilk ***@***.***> wrote: @davidjmcclelland <https://github.com/davidjmcclelland> is this code open source, by any chance? (I'm not promising anything, but I've been on a bit of a JS memory leak bent.) And is this browser specific? That sounds odd. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#870 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AANwnUCrRdNMSVMXy5hdziOeYIgeyFjtks5rtZdVgaJpZM4CAi9z> .