On Nov 27, 2007, at 18:04, Chris Bowditch wrote:
Andreas L Delmelle wrote:
<snip/>
Sorry, hadn't updated for over a week... :/
No problem. It was easy enough to revert to the previous revision.
<snip/>
Ouch! I wasn't using the right expression to map from the stale
entry's hashCode to the corresponding bucket index.
Fixed in the diff in attach (which removes some more obsolete code
as well, and adds one more layer of synchronization around the
loop where the votes for rehash are counted)
Bad news I'm afraid. NPE after 25 docs. This time in the while loop
condition in cleanSegments:
Cannot reproduce it here...
java.lang.NullPointerException
at org.apache.fop.fo.properties.PropertyCache.cleanSegment
(PropertyCache.java:99)
... and this line can normally not cause an NPE.
Which JVM did you test it on? Do you run multiple concurrent threads,
or is it just a plain single-threaded iteration?
<snip />
Also, I'm not sure why synchronization is needed if the cleaner
threads have been removed?
Because the caches are shared between multiple FOP sessions in the
same VM.
Cheers
Andreas
