I runned the App1Test1 in 100 iterations and 10 sessions with then new wicket object outputstream against the normal one
the page size for this test drops from 32KB to 18KB Also the test runs +20% quicker then with the normal one. It runs slower then yesterday currently because i do a lot more synching so that i got everything right Now we have to see if we can drop the synching again as much as possible. The WicketObjectOut and In are really only reliable for one jvm instance! It is not meant for transfering over the wire and can't be used for restarts. For that to work i need to write full classnames to the stream instead of just a short.. So i guess that will make it bigger again but we can do it if needed. Matej can you test if the current code still reports errors? johan 4 runs with wicket serialisation: ******** finished in 86313 miliseconds Total time in serialization: 60281 Total time in saving: 71571 (22ms per page) Bytes saved: 57709624 (18KB per page) Pages saved: 3196 ******** finished in 90672 miliseconds Total time in serialization: 65338 Total time in saving: 77732 (28ms per page) Bytes saved: 49858372 (18KB per page) Pages saved: 2756 ******** finished in 92391 miliseconds Total time in saving: 78728 Total time in serialization: 87231 ******** finished in 91390 miliseconds Total time in saving: 71092 Total time in serialization: 94432 4 runs with object serialization: ******** finished in 144390 miliseconds Total time in serialization: 145277 Total time in saving: 111512 (35ms per page) Bytes saved: 103719333 (32KB per page) Pages saved: 3203 ******** finished in 125953 miliseconds Total time in serialization: 148461 Total time in saving: 94538 (25ms per page) Bytes saved: 123523671 (32KB per page) Pages saved: 3818 ******** finished in 116813 miliseconds Total time in saving: 87849 Total time in serialization: 140883 ******** finished in 133860 miliseconds Total time in saving: 98868 Total time in serialization: 175411
