Is there any plan to bump the netty to version 4? I know it includes many API changes.
Yep, it's on the roadmap. We work closely with the netty team, and hope to be able to spend some time to get it done soon. In the mean time, if you want to try it out yourself, there has been a lot of work done recently to decouple finagle from netty 3. I would suggest taking a look at DefaultClient and DefaultServer.
Thanks for your fast reply. I'm looking forward to the new version. I will see the suggested codes.
Do you want to close this issue now or when the migration is done?
In Twitter Engineering Blog there's a post from October 2013 by Trustin Lee (Netty project maintainer and Twitter employee) saying that they successfully migrated to Netty4.
Now, 7 months later, Finagle is still based on Netty 3.
I'm not sure if the open source version is lagging behind or Twitter decided to roll back for some reason.
We did not say we successfully migrated to Netty 4. Migration is in progress.
Sorry, my bad, the article only said the migration was in progress. In fact a more recent article says Finagle needed (needs?) some refactoring before upgrading to Netty 4. I will definitely stay tuned
No worries. Thanks for staying tuned :-)
Is there a rough ETA?
My wild guess is this Christmas. It's a little bit more complicated because Finagle exposes Netty 3 types in its public API. Complex migration path is ahead of us.
Note: there's a DoS vulnerabliity in Netty through version 4.0.19...
@ernieKovak thanks for the heads up. Twitter keeps track of vulnerabilities that appear in any lib we depend upon, and make sure to upgrade if there's a problem. Happily, finagle doesn't use any of the netty websockets stuff, and we're on netty 3.9.4, so the version of netty we're on has had the fix for a long time. When we upgrade to netty 4, we'll upgrade to the latest version, so we won't have to worry about bugs in netty 4.0.19.
Christmas is approaching @trustin :)
Hi @abbaspour—see this Finagle blog post for a much more up-to-date and detailed outline of the upgrade plan.
Any news on this subject?
@note no new developments, work is still actively underway internally but we've intentionally taken the longer path of removing netty types from our protocols' public api to make future upgrades easier.
That's how Google's gRPC avoids exposing netty in the public API.