james-server-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Robert Burrell Donkin" <robertburrelldon...@gmail.com>
Subject [mime4j] MultiReferenceStorage Locking
Date Sun, 30 Nov 2008 22:59:13 GMT
uses method based synchronization to protect the reference counting
variable. i wonder whether this is the best approach. (i tend to
prefer explicit monitors that cannot leak for defensive reasons.)

in particular

    public synchronized void delete() {
        if (referenceCounter == 0)


        if (referenceCounter == 0) {

1. storage is an interface and so storage.delete() represents a leak
of the thread of execution. i wonder whether it is possible for
competing storage deletes to produce a deadlock.
2. storage.delete() may be a slow operation. the locking appears to be
protecting only the reference counting. if so then the delete may not
need to be synchronized,

it might be possible to avoid explicit synchronization by using
AtomicInteger instead


- robert

To unsubscribe, e-mail: server-dev-unsubscribe@james.apache.org
For additional commands, e-mail: server-dev-help@james.apache.org

View raw message