Hi,
If you ask me, not serializing the final field was more a bug than a
feature. (I know, GWT RPC is not the same as Java serialization)
Do I need to enable something to get warnings on final fields not being
serialized in 2.5.1 or is that standard behaviour.
I am a very active user of the
I am very much in favor of moving things forward with breaking changes as
long as we can see most of the damage up front.
Let me check internally if we can find out if Johns concerns are valid so
we might be able to implement this much simpler.
On Thu, Jun 6, 2013 at 9:05 AM, David
On Wed, Jun 5, 2013 at 7:46 PM, Stephen Haberman
step...@exigencecorp.comwrote:
I don't have a good idea about what could break
I believe nothing would break--it's that final fields that were
previously not going over the wire would now go over the wire.
It's also possible that the newly
Let me check internally if we can find out if Johns concerns are
valid so we might be able to implement this much simpler.
Cool; if having a patch to run through the Google submit/test queue
thing would help, let me know and I can put something together.
- Stephen
--
Hey,
So, there is a CL out there that I have kinda/sorta been shepherding
along to add support to GWT-RPC for final fields.
The current CL is kinda complicated right now, because we have at least
two flags about whether to turn final-fields on/off and also whether to
warn about it when it is
I don't have a good idea about what could break, but who knows what we'll
find. I think we will at least want a flag to turn it on and off. It might
be temporary just to land the patch; not sure if it should survive until
the 2.6 release.
On Wed, Jun 5, 2013 at 9:03 AM, Stephen Haberman
I don't have a good idea about what could break
I believe nothing would break--it's that final fields that were
previously not going over the wire would now go over the wire.
So, I guess technically that is a change in semantics--so it wouldn't
be like oh, your GWT app fails to compile, more
On Wed, Jun 5, 2013 at 10:46 PM, Stephen Haberman
step...@exigencecorp.comwrote:
I don't have a good idea about what could break
I believe nothing would break--it's that final fields that were
previously not going over the wire would now go over the wire.
So, I guess technically that is a
Does having the flag really make it that much more complicated?
It would not seem like it, but yes. :-)
Specifically passing the context down through the layers to the right
spot to say do I serialize final fields? took a bit--one of the first
patches that went by used a static field to avoid