@Mike ... by "deliberately" I meant differently than in previous versions of jQuery, which were more forgiving. There are links to relevant texts, I am sure ... maybe this : http://www.learningjquery.com/2009/03/43439-reasons-to-use-append-correctly @Phaedra ... thanks for not re-acting in anger ;o) After we remove all the obvious obstacles and have a valid code we can concentrate on leaks that are result of the library, despite the fact the code is right. jQ can not work properly in the presence of deliberate (or accidental) mistakes in the code that uses it. And valid html too.
Otherwise, I would like to know how are other browsers behaving? It seems IE7 has singled out itself in this case ;o) On Sep 23, 8:57 am, Phaedra <phaedra.casa...@gmail.com> wrote: > > Phaedra, how are you identifying the number of leaked nodes? > > I'm using sieve 0.08. I take note of the used nodes in the home page, > then, i open a dialog (jquery), again take note of the nodes. > After closing it they are always there, removed in the dom but > always there in memory. These nodes are free fom events. > > Same in the two examples, a blank page with a button with a onclick > event. > > I can accept that the clock example was not well formed, but this was > intentionally, apart of caching the selector. > Changing the html inside the div call the "empty" function of jquery, > that does nothing about memory leaks. > So, i agree that .text() was better, but my example was about memory > leaks, not performance. When ie7 number of used nodes grow, it slow > down, making the spa too slow.
