Closed GoogleCodeExporter closed 9 years ago
this happen with iOS.newVM=true
Original comment by rbry...@gmail.com
on 14 Jan 2015 at 8:08
We made improvements to the GC which should hopefully fix that
Original comment by shai.almog
on 18 Jan 2015 at 2:10
hi shai,
my app still crash, I didn't check well if the error is the same but the
procedure is the same.
Original comment by rbry...@gmail.com
on 28 Jan 2015 at 6:29
Try using the TreeSet constructor that takes a comparator as one of the
parameters. (And, of course implement a Comparator the does the comparison).
Let me know if that works. I think I know what may be happening here.
Original comment by st...@weblite.ca
on 28 Jan 2015 at 11:55
Just tried running this test case and it fails with the same error. Screenshot
attached or Xcode stack and error.
Running through a few things. Likely a GC problem though.
Original comment by steve.ha...@codenameone.com
on 30 Jan 2015 at 10:16
Attachments:
Original comment by steve.ha...@codenameone.com
on 30 Jan 2015 at 10:20
Also I note that changing to providing explicit comparator for treeset makes no
difference. My test case is attached.
Original comment by steve.ha...@codenameone.com
on 30 Jan 2015 at 10:23
Attachments:
thank you steve, I didn't have the time to do these checks by myself, I'm
sorry...
Original comment by rbry...@gmail.com
on 30 Jan 2015 at 11:36
I just tried Steve's test case on my device and it works fine. No idea what
sort of crash you are still seeing.
Original comment by shai.almog
on 31 Jan 2015 at 11:34
After updating to the latest changes this morning, my test case (with the
Comparator in the constructor of TreeSet) now passes. If I use the default
constructor, I still get the same error.
I suspect this may be related to the fact that String doesn't implement
Comparable in the NewVM's JavaAPI. I have already created workarounds for all
of the Number classes. I'll look into this more on Monday to see if I need to
do something similar with Strings.
Attached is a new test case (without the Comparator in the constructor).
Original comment by steve.ha...@codenameone.com
on 31 Jan 2015 at 2:03
Attachments:
I'll just make String implement comparable.
Original comment by shai.almog
on 31 Jan 2015 at 3:53
It would be interesting to see if this works on J2ME as their String class
doesn't implement Comparable either. My guess is that it will throw a
ClassCastException.
Original comment by steve.ha...@codenameone.com
on 31 Jan 2015 at 4:50
sorry for the delay but is a busy period.
Unfortunately I still have problems with TreeSet. Just to be sure I tried to
build a new form with a Tree that shows integer as branch ending with leaf
having 4 digits: all was ok until I used a TreeSet<Integer> to collect children
values before transferring them to a Vector<Integer>. I tried to notify another
issue but it has been refused by shai saying that it is a VM break down.
As you can see it is not a matter of String because Integer lead to the same
result.
Try to use the attached BuggyTreeForm.
Please, I'm near to publish for iOS, they just disapproved my binary because of
oldVM(which runs perfectly) being only 32bit.
Original comment by rbry...@gmail.com
on 1 Mar 2015 at 1:11
Attachments:
I just built your test case and ran it on iPhone 4S/8.1, and it runs fine, so
I'm guessing Shai already fixed this.
Original comment by st...@weblite.ca
on 2 Mar 2015 at 4:31
Yes, I got a good test case today and committed a fix this morning. It seems
that TreeSet implements a hierarchy of interfaces in a rather unique way that
caused one of the interfaces in the hierarchy to map method offsets badly.
Original comment by shai.almog
on 2 Mar 2015 at 6:33
Original issue reported on code.google.com by
rbry...@gmail.com
on 14 Jan 2015 at 7:37