Hi all
I would say we can also see mutability/immutability as simply two options..
In general it is a decision of the event/payload designer, If the event is
mutable or not, and if it is mutable, it should be implemented in a thread
safe way with (hopefully) minimal locking/synch efforts. To mention the
ladder somewhere in the spec might be useful.
I would hereby tend to follow the same policy for synch and asynch events
(you never know what else payload consumers do with the event). I think it
is sufficient to warn users that mutable asynch events basically can create
race conditions and locks if thread-safety is not given, but I would not
generally disable/disallow it.
-Anatole
2014-12-17 17:49 GMT+01:00 Antoine Sabot-Durand <antoine(a)sabot-durand.net>:
Ok guys,
Let’s do it again. I didn’t say we have to forbid the mutability I said we
have at least to explicitly write that it’s mutable and seriously think of
having it immutable for event fired asynchronously.
(Pete) I don’t think it’s specified. As objects are, by default in Java,
mutable, I would assume that payloads are implicitly mutable.
Sorry @Pete I don’t agree with your point. Yes, in Java object are mutable
but firing an event is not a standard Java feature : you send your object
to a black box and let this box dispatch your object to listeners
transforming one call to multiple call : it’s far from standard Java rules.
Even if it’s not written it’s an observer pattern and there are people out
there thinking that introducing mutability in observer is an anti-pattern
since some listener will receive a different payload than the one that was
sent to them.
It’s like making a method call and having no guarantee that the parameter
received in the callee has the same value that in the caller...
I won’t start discussion on bad practice or anti pattern as I also use
mutability in event but there as much reason for user to assume their
payload will be mutable than the other way around.
I can assure you that when I give a talk on CDI, this payload mutability
is often a surprise for attendees...
(Romain) why isn't it portable?
So yes @Romain it’s not portable (in theory of course, since both
implementations support mutability). Someone could write a CDI
implementation with event payload immutability without any issue with the
spec and TCK.
Most of you are so dependent of this feature that you only reacted to the
idea or forbidding it (which wasn’t the content of my mail) ;). So we all
agree that it’s an important feature. Therefore what’s the issue to specify
this mutability and add TCK test for it ?
Now I don’t deal with that subject for nothing, we are planning to
introduce Async events. I think that it’ll bring extra complexity if we
support mutability in async events. And even if I’m wrong and we finally go
for mutability in async events, this will lead to possible side effect
(lock) that could have impact on perf, so it should be explicitly written
IMO.
Antoine
_______________________________________________
cdi-dev mailing list
cdi-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/cdi-dev
Note that for all code provided on this list, the provider licenses the
code under the Apache License, Version 2 (
http://www.apache.org/licenses/LICENSE-2.0.html). For all other ideas
provided on this list, the provider waives all patent and other
intellectual property rights inherent in such information.
--
*Anatole Tresch*
Java Engineer & Architect, JSR Spec Lead
Glärnischweg 10
CH - 8620 Wetzikon
*Switzerland, Europe Zurich, GMT+1*
*Twitter: @atsticks*
*Blogs: **http://javaremarkables.blogspot.ch/
<
http://javaremarkables.blogspot.ch/>*
*Google: atsticksMobile +41-76 344 62 79*