Re: Serialisation of Object/BitSet

2009-04-11 Thread Vitali Lovich
I have no experience with the JDO stuff, so I can't comment on it. On Sat, Apr 11, 2009 at 5:33 AM, Dean S. Jones wrote: > > Every time I run into this kind of issue, BitSet, BigInteger, > BigDouble I dig into the source code and find one common element: > long, long, long. > How big exactl

Re: Serialisation of Object/BitSet

2009-04-11 Thread Dean S. Jones
Every time I run into this kind of issue, BitSet, BigInteger, BigDouble I dig into the source code and find one common element: long, long, long. 3 GWT projects I thought I was smarter than the GWT team and implemented the emul classes and serializers. Worked most of the time, but failed on t

Re: Serialisation of Object/BitSet

2009-04-11 Thread datanucleus
I'll move discussion to the other thread that Vitali linked to avoid duping. Thx for your time. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Google Web Toolkit" group. To post to this group, send email to Google-W

Re: Serialisation of Object/BitSet

2009-04-11 Thread datanucleus
Well not being a GWT user I can't comment on the details of what is going on in the GWT side, but from the enhanced class side here you can see an enhanced class http://db.apache.org/jdo/enhancement.html with the before and after. The field "jdoDetachedState" is the one referred to, and this relie

Re: Serialisation of Object/BitSet

2009-04-11 Thread Vitali Lovich
On Sat, Apr 11, 2009 at 5:02 AM, Dean S. Jones wrote: > > At this point, I'm not going to bet my life on GWT simulation of long > and bit operations. I would because first of all you can check the code - this kind of stuff has been written hundreds of times for all sorts of different bit sizes.

Re: Serialisation of Object/BitSet

2009-04-11 Thread Dean S. Jones
At this point, I'm not going to bet my life on GWT simulation of long and bit operations. boolean[] is less efficient, but easier to trust. --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Google Web Toolkit" group.

Re: Serialisation of Object/BitSet

2009-04-11 Thread Vitali Lovich
Are you sure the GWT long acts differently than a true Java long? My understanding was that as of 1.5 the long on the client correctly emulates a real long primitive. No? http://groups.google.com/group/Google-Web-Toolkit-Contributors/browse_thread/thread/3c768d8d33bfb1dcseems to be the thread th

Re: Serialisation of Object/BitSet

2009-04-11 Thread datanucleus
Thx Dean. So if this "Object[]" field had Object[2] stored as boolean [] and Object[3] stored as boolean[] then this would work ? --~--~-~--~~~---~--~~ You received this message because you are subscribed to the Google Groups "Google Web Toolkit" group. To post to

Re: Serialisation of Object/BitSet

2009-04-11 Thread Dean S. Jones
If you look at the Java source for BitSet, you'll see it uses an array of long[] JavaScript has no concept of long, so GWT has an "simulation" of long, but it's not perfect. One could easily write an emulation class and serializers, but it's not likely the bit math would be correct without to

Serialisation of Object/BitSet

2009-04-11 Thread datanucleus
Hi, I'm not a GWT user, instead I write DataNucleus (used by AppEngine). There's an issue around GWT and RPC serialisation when using a bytecode enhanced class. The class (when enhanced) has a field Object[] jdoDetachedState; which basically contains Object[0] is an object of the primary key ty