Hi Matthias - > well, the JCP (true for 99% of the specs) is slow; > and I am sure, the EG will find some reasons to not make changes... > :-)
In that case here is an alternate approach: 1. Keep using the application map. 2. Log bugs against any servlet implementations that are stuck in Hashtable-land (eg. Glassfish) as this implementation poses scalability problems and there is an obvious and trivial solution (ConcurrentHashMap). 3. Not worry about the possibility of servlet implementations that are providing totally unsynchronized access to ServletContext properties. The reason why I am not worried about #3 is because I don't think these exist. We can also do: 4. Follow up with the JCP for clarification. Andy > > sorry, but is my personal feeling. > > -M > >
