4:55 p.m.
Hi,
I've noticed that if I do something stupid in one of my bean's
constructors (a NPE for example ;) the exception is being buried by
SimpleBean.create which has 'trys' and 'finallys' but no catch. I assume
there's a good reason for not rethrowing the exception here (what is it
by the way?), but could we maybe log it?
Cheers,
Pete.
5:51 p.m.
A try-finally doesn't swallow exceptions, and a catch-rethrow is not
necessary unless you need to inspect the exception somehow (e.g. to
provide more context) and swallow-log is a horrible approach.
I suspect you have some other problem, probably inside the logic which
rethrows exceptions invoking constructors as CreationExceptions.
On 24 Mar 2009, at 21:55, Peter Royle wrote:
--
Pete Muir
http://www.seamframework.org
http://in.relation.to/Bloggers/Pete
2:03 p.m.
A try-finally doesn't swallow exceptions,
That's what I thought! But all bets are off when the finally block
throws an uncaught exception :).
and a catch-rethrow is not necessary unless you need to inspect the
exception somehow (e.g. to
provide more context) and swallow-log is a horrible approach.
Agreed :)
I suspect you have some other problem, probably inside the logic
which
rethrows exceptions invoking constructors as CreationExceptions.
The problem in this case is that an NPE is thrown in the finally block
because it relies on a variable which is initialiesd in the try block
which bailed out. This new NPE causes the original exception to be
buried.
The fix is easy though - I've jsut made DependentStorageRequest capable
of comparing against null.
http://fisheye.jboss.com/changelog/Webbeans/?cs=2254
Cheers,
Pete.
2:41 p.m.
Actually this is probably the safer way to do it:
http://fisheye.jboss.com/changelog/Webbeans/?cs=2255.
The previous commit changed the behaviour in a couple of subtle ways
which is not a particularly good thing.
Pete.
On Sun, 2009-03-29 at 19:03 +0000, Peter Royle wrote:
6:33 p.m.
On 29 Mar 2009, at 20:41, Peter Royle wrote:
Explain - I can't see how this is currently safe given that you will
get a CCE if I do
dependentStorageRequest.equals(new String());
Anyway, I made this use the standard equals idiom and the test suite
appears unaffected...
Yes, so (as you say) we make it so we don't throw an exception there :-)
--
Pete Muir
http://www.seamframework.org
http://in.relation.to/Bloggers/Pete
4:39 p.m.
OK good points :). By 'safe' I just meant less change to the behaviour
of the equals method so less likely to break any existing functionality
- even though that would have been dead easy to check. Cautious, yet
lazy :).
But you're right - eliminating the CCE as well is the right thing to do
here.
On Mon, 2009-03-30 at 00:33 +0100, Pete Muir wrote:
6020
days inactive
6026
days old
5 comments
3 participants
participants (3)
-
Pete Muir -
Peter Royle -
Peter Royle