Hi,
I¹ve been crawling through an OutOfMemory dump today and traced a problem
back to the json.org parser.
Apparently this truncated JSON will create an 8 million+ size ArrayList:
mySource java.lang.String
"[{"type":"FETCH_PEOPLE","idSpec":"OWNER","profileDetail":["id","name","thum
bnailUrl"],"sortOrder":"topFriends","filter":"all","first":0,"max":20},"
What do people think about switching over to json-lib
(http://json-lib.sourceforge.net/) instead? It appears that this is already
used by Abdera and some other Apache projects.
In addition to this problem I¹ve already had to remove the code that checks
the Mandatory annotation. That code ends up checking a synchronized
HashMap, which causes severe contention problems under load.
Paul
[EMAIL PROTECTED]
