Firefox 28 crashes with d3.js
Crash id: bp-bf42e5a7-54f6-46a9-aeb9-8675f2140402 (edit added bp- forum then converts to link ~J99)
An out of memory situation causes a crash in Firefox Browser rel 28 when using d3.js visualization library and applying path tweening in a transition to adapt a SVG path. "uncaught exception: out of memory" console message is displayed.
This seems to be a regression issue: The example code that causes the issue now in rel 28 worked well in rel 27. Also, the example runs without issues in Google Chrome v33.
The original example code that runs without issues in both rels 27 and 28 can be found in: http://bl.ocks.org/mbostock/3916621
A slightly modified version can be found in https://gist.github.com/mmakipaa/9951985 This code runs without issues rel 27, but causes an out of memory crash in rel 28. Compared to the original example, path string variables d0 and d1 have been replaced with more complicated paths and SVG path element transformation is commented out.
Izmjenjeno od strane John99
All Replies (4)
I tested on windows 8 and mac os x, the one on win 8 looks much smoother. One is not crashed though when I turned hardware acceleration off. The memory available in the crash report is about 3.5 gb and virtual 4294836224
There is high memory usage: https://bugzilla.mozilla.org/show_bug.cgi?id=943985
search for morehttps://bugzilla.mozilla.org/buglist.cgi?query_format=specific&or...
Does the memtest help? http://www.kelleycomputing.net/rember/
Filing a bug for this with the info above will be a pretty good bug description in [bugzilla.mozilla.org] too.
I have just managed to test this on Linux & Windows 7. Linux has no problems. Windows does not crash but does not animate and throws an error even with hardware acceleration disabled. I only had the tab affected the browser continued to behave, so I did not look at about:memory.
I not understand code but as you do I wonder if you could create a reduced code test case to demonstrate the problem. Would Firefox's debugger help you ?
Presumably if you need to compare with Firefox 27 you have no problems running multiple installs with separate profiles.
Note a slightly related thread /questions/992803
Thanks for answers and testing.
I tried with the latest nightly build and the code seems to run without issues, do you think it still would make sense to report this as an issue in bugzilla?
Firefox 29 beta and later work for me on Linux.
The current release (28) gives a weird pop-up (no text) that covers the entire screen width and that I can close with the ESC key (no animation).