Message2355
Logged In: YES
user_id=971697
The analysis provided is a dynamic one. Therefore, it only
shows those problems that it actually encounters. Thus
there may be additional problems that have not yet been
uncovered.
I believe that it must provide false positives, because of the
example you presented (in #2). I could not figure out how
nameindex could be causing a problem in that situation!
The reason I used a class level lock was mainly defensive
programming. I initially tried providing a separate lock for
each of the member variables, but this led to very
complicated locking schemes due to the fact that the
variables are used in a lot of different methods. Not knowing
enough about how the methods interact with each other, I
generally ended up causing deadlocks in the PyJavaClass. So
I went back to a class lock.
Your suggestion of locking the proxyClass may work. I do not
want to claim that my solution is the only one. It is just what
I was able to come up with.
Finally, the code is from the 2.1 version, and not the latest
version from CVS. I checked the CVS to see if anything had
been modified in the PyJavaClass that may affect the
threading because if it had already been fixed, I did not want
to fix it again. But I couldn't find any fixes that dealt with
the threading problems I saw. |
|
| Date |
User |
Action |
Args |
| 2008-02-20 17:18:23 | admin | link | issue898709 messages |
| 2008-02-20 17:18:23 | admin | create | |
|