Externally destroying a DataTable can lead to a memory leak

[DataTables]

I've recently been looking into a memory leak issue for a DataTables user who makes use of your handy library (thanks for providing it!).

What appears to be happening is that the DataTable is being destroyed use the DataTables destroy() method, but then the data source that has been powering the DataTable data (from an ng-repeat) is updated. That is causing the DataTable to be reinitialised by this library and thus a leak occurs.

I've put an example together showing the issue. To reproduce, just start a profile in Chrome and then run the run() method in the console.

I've also put an example together that does not modify the data source after the DataTable has been destroyed and does not exhibit this issue.

As a possible fix, perhaps it would be possible to listen for the destroy.dt event from DataTables and when found, stop performing any updates to the DataTable. This would require an external destroy event to be distinguished from an internal one, as I see you use the destroy at lot internally.

[l-lin]

I see. Thanks for reporting it 👍

[l-lin]

I think it's resolved with #326.
As I'm quite new with profiling JS codes, can you point out where the leak is in the chrome profiling manager?
Because I can't see the difference between the one with the memory leak and the one without.

Using Chrome timeline, with leak:

Without leak:

Is it the number of listeners?

Thanks!

[deifos]

Any updates on this issue?

[l-lin]

Unfortunately, as Angular 2 is coming out, I lack the motivation to find the fix for this peculiar bug for the versions 0.X.X of angular-datatables... If some kind soul is willing to fix this... 😖

[l-lin]

Closing as issue is too old and the version is no longer maintained.