Thurs July 16 2020 GWT community call

Skip to first unread message

Colin Alworth

Jul 15, 2020, 5:02:08 PM7/15/20
to GWT Contributors
We have a shorter itinerary this week - I'll record a short piece on j2cl-maven-plugin and how to start a project with it, try using pieces from the ecosystem.

Dmitrii, Ahmad and I will continue our brainstorming about efficiently producing both optimized output and minimizing the work the compiler does for i18n/cldr generated code.

We'll probably also continue poking around j2cl/closure output to recap what we learned about string concatenation last time, understand why this makes sense, possibly try to make the output improve a bit.

The first video from last call was published earlier this week, the second one will probably go out tomorrow. The first video can be found at, we'll continue to publish the other videos to the same channel.

We'll get started at our usual time, 5pm CEST / 11am EDT. The call will be on Google Meet again, but we're open to options that can record at a higher resolution and not force the presenter's picture into frame and hurt readability.

Michael Conrad

Jul 21, 2020, 9:24:13 AM7/21/20
How did it go?

Colin Alworth

Jul 23, 2020, 9:48:24 AM7/23/20
to 'Goktug Gokdogan' via GWT Contributors
It went well - it was a smaller crowd, probably in part due to fewer "talks" and more "contrib", but that's a balance we're going to have to continue to strike.

The main focus was on getting closure to provide the sort of optimization we've come to expect from gwt2, and we had a surprisingly large amount of difficulty in getting it to do this. I'm doing a writeup now for the closure compiler list, since we've got a fairly simple set of plain example JS that demonstrates the issue. The biggest concern is that this means that you have to use raw JS to "explain" this pattern to the compiler, otherwise it will not understand and will not bother to prune/optimize code, but our guess right now is that this is a specific enough of a use case that we can provide tooling to solve this problem, and that tooling can generate JS instead of developers writing it. Also worth noting that j2cl+closure has been at least on par with gwt2 if not smaller in the examples we've tried so far, so especially if it tends to generate sub-optimal code in a few cases, we know it is doing a far better job than gwt2 in other cases.

Next week we have two talks planned already, and have room for 1-2 more if anyone is interested. As usual, the call will start an hour early with no set topic, instead anyone can ask questions or make suggestions on what we can discuss.

  Colin Alworth

You received this message because you are subscribed to the Google Groups "GWT Contributors" group.
To unsubscribe from this group and stop receiving emails from it, send an email to

Reply all
Reply to author
0 new messages