Closed GoogleCodeExporter closed 9 years ago
Issue 75 has been merged into this issue.
Original comment by peter.ry...@gmail.com
on 12 Sep 2012 at 4:57
See a related issue http://code.google.com/p/chromedevtools/issues/detail?id=77
Original comment by peter.ry...@gmail.com
on 12 Sep 2012 at 4:58
I think, since the "strings" are parsed as [numbers], the object might get
parsed as array, and the few billion indices between the numbers get assigned
'undefined' automatically, as this is how arrays internally work (and was the
reason for using objects in stead of arrays in the first place, which lead to
my original bugreport).
Original comment by redsandro
on 12 Sep 2012 at 7:18
Sander,
could you please expand your last comment. Particularly, what "parsed as array"
means.
Peter
Original comment by peter.ry...@gmail.com
on 12 Sep 2012 at 9:00
I have seen this behavior in a NodeJS app I am working on. NodeJS is build on
V8. In NodeJS/javascript, when you define an array with two big integers as key
like so:
var ar = [];
ar[1000000000] = 1;
ar[3302611133] = 3;
internally, all keys in between are set in memory like so:
ar[1000000000] = 1;
ar[1000000001] = undefined;
// ...
ar[3302611132] = undefined;
ar[3302611133] = 3;
You won't see this in the Variables Explorer, but memory will fill and the V8
(nodejs) application will crash.
This looks a lot like what happens if you use an {object} in the way described
by this bug, as long as you use numbers (as string) that are parsed as integers
like in bug #77.
Except, this bug with the object only happens in the Variables Explorer or the
Expressions View in Eclipse, not in the actual NodeJS program. Hence, it feels
like - on the devtools side of things - the object is internally somehow
converted to array which causes the crash.
I hope this makes more sense.
Original comment by redsandro
on 12 Sep 2012 at 9:47
As if the Variables panel or Expressions panel run out of reserved memory and
crash, just like the app does with arrays.
Original comment by redsandro
on 12 Sep 2012 at 9:49
Oh, I see. It's a good guess, but in fact it's unrelated. It's just a slightly
inaccurate code in Debugger, the fix is ready and soon will be landed:
http://codereview.chromium.org/10913231/
Original comment by peter.ry...@gmail.com
on 13 Sep 2012 at 2:29
Nice, thanks.
Original comment by redsandro
on 13 Sep 2012 at 8:04
Fixed in HEAD
Original comment by peter.ry...@gmail.com
on 19 Sep 2012 at 4:09
Fixed in 0.3.8
Original comment by peter.ry...@gmail.com
on 4 Oct 2012 at 8:33
Nice!
Original comment by redsandro
on 4 Oct 2012 at 9:10
Original issue reported on code.google.com by
peter.ry...@gmail.com
on 12 Sep 2012 at 4:42