NEWHPROF: RFE: Object age and ability to determine sort order
This doesn't have to go into hprof necessarily, but something that jfluid
does would be handy to try. Count the GC's with GC_FINISH event, and add
the current GC number to object creation, and/or create an average age
of all objects allocated at a site. Then include this data in the summary.
In other words, provide some more help in finding leaks by providing more
data about the objects. Ideally, the tool could perhaps list the most likely
objects that could be leaks? Just a thought.
Also, probably worthwhile to allow different sorts of the data, it's
pretty fixed right now.
Now perhaps what should happen is that the output should be done in XML
or some basic table form, and some kind of viewing tool could be used to
view that output, selecting the sort criteria after the collection.
Perhaps at runtime with the data sent down a socket or transport mechanism.
just ideas I didn't want to forget about.
###@###.### 2004-07-23
This doesn't have to go into hprof necessarily, but something that jfluid
does would be handy to try. Count the GC's with GC_FINISH event, and add
the current GC number to object creation, and/or create an average age
of all objects allocated at a site. Then include this data in the summary.
In other words, provide some more help in finding leaks by providing more
data about the objects. Ideally, the tool could perhaps list the most likely
objects that could be leaks? Just a thought.
Also, probably worthwhile to allow different sorts of the data, it's
pretty fixed right now.
Now perhaps what should happen is that the output should be done in XML
or some basic table form, and some kind of viewing tool could be used to
view that output, selecting the sort criteria after the collection.
Perhaps at runtime with the data sent down a socket or transport mechanism.
just ideas I didn't want to forget about.
###@###.### 2004-07-23
- relates to
-
JDK-6239651 New binary heap dump format (to replace hprof format=b)
-
- Closed
-