Calling directory.close() before directory.getChildren() helps indeed. But it isn't nice at all especially because this is the default behavior. It
will definitely require addressing.

I think a distinction should be made between short runs (e.g. Ant task) and long runs (like a file manager) for VFS. By default a long running task should be considered as happening (so no caching). If a client knows that it needs a short running task then it should configure VFS accordingly. However the default setting should provide a real time view of the file systems.

I don't know, but shouldn't this behavior considered as an (architectural)
bug?

Well, I've thought the same thing for a while.
But not using caching at all will have quite an
impact on speed. So turning it off completely by
default ...don't know if that makes sense on the
protocol level. You might want to have something
in between.

We ended up just properly closing everything which
made all problems disappear - for us.

My 2 cents

cheers
--
Torsten

Attachment: PGP.sig
Description: This is a digitally signed message part

Reply via email to