303474 - Memory leak when appending/removing elements to/from a document

Status: RESOLVED INVALID

(Core :: DOM: Core & HTML, defect)

Description

[Julien Lecomte]

User-Agent:       Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4322)
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.8) Gecko/20050511 Firefox/1.0.4

Attached is a test case. Press the 'start test' button and watch the memory 
increasing, slowly first, then very steadily.

Reproducible: Always



Expected Results:  
Since the containing element is removed from the DOM at each iteration, the 
memory footprint should stay approximately the same.

Comment 1

[Julien Lecomte]

Attachment: Test case

On my machine, the memory footprint stays approximately the same for about 30
seconds, then it starts increasing steadily.

Comment 2

[Julien Lecomte]

Just tried on Firefox 1.0.6 (User agent: Mozilla/5.0 (Windows; U; Windows NT 
5.1; en-US; rv:1.7.10) Gecko/20050716 Firefox/1.0.6) on a different Windows 
machine, and I get the same result (memory footprint flat for 30/40 seconds, 
then steadily increasing)

Comment 3

[Julien Lecomte]

Deer Park Alpha 2 (User agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; 
rv:1.8b3) Gecko/20050712 Firefox/1.0+) has the same exact problem.

In comparison, IE6 memory footprint is very stable.

Comment 4

[Julien Lecomte]

Attachment: Memory profile graph

Blue: Firefox 1.0.6
Pink: IE6 SP2 for comparison

You can observe very clearly the leak on this graph after only 20 seconds.

Comment 5

[Julien Lecomte]

Attachment: Fixed test case

There was a small typo in the first test case. This new test case fixes it. It
does not change anything to the result shown in the graph.

Comment 6

[Julien Lecomte]

Attachment: Memory profile graph (2)

After removing the loop from the test case, the same thing happens, but later
(85 seconds instead of 20 seconds)

Comment 7

[Julien Lecomte]

Attachment: No leak graph

Ran the same test for 10 minutes and the GC finally kicks in...

Comment 8

[Julien Lecomte]

Closing invalid per the last experiment.