Elemental 2 and J2CL timeline

[Paul Stockley]

Does anyone have any idea when elemental 2 will be available? Also what is a rough timeline for a version on J2CL being available, is it 6 months, a year or more away? 

The reason I ask is that I am thinking about how our company can migrate to the new compiler. We will need widgets and UiBinder for quite a while, even though we will slowly migrate to React. I was contemplating creating a project to port the basic Widget framework to use something like elemental. Also I was thinking about how to build a new version of UiBinder that didn't use GWT.create.

[Colin Alworth]

I can't speak for J2CL's timelines, as it is an internal project at this time. However, it has been stated that it will support GWT's JRE emulation and JsInterop out of the box, and the expectation is that it will simply be a transpiler and won't support generators (suggested that projects move to APT or the like), or linkers (since the Closure Compiler manages this part of things).

By itself, this means that Widget can work, though Element will need to be reimplemented in JsInterop since JSOs won't exist (and JSNI is likely to have a replacement as well). On the other hand, UiBinder uses the GWT Generator system, which doesn't at this time have a clear upgrade path.

To think about life without GWT.create(Foo.class), consider APT-based projects like AutoValue, where class sources are generated on your classpath (by your build tool, and your IDE if configured correctly) so that you can reference these generated classes directly from within your sources. For cases where you want to pick the right implementation (rtl, locale, device formfactor, etc), a factory methods can also be declared within the generated code, and you can reference  that rather than a constructor.

[Paul Stockley]

I am assuming that Google isn't really interested in Widgets and UiBinder for J2CL given their usage of GWT in inbox etc. I could be wrong. Assuming I am not, I would like to investigate starting up some community projects for Widget and Uibinder support. These won't be trivial to replace so we will need quite a long lead time. However, I think if we can show the community that they will be able to rely on these in the future, it will go a long way to reducing the fear that GWT is dead. 

[David Becker]

As a GXT developer at Sencha, we are keenly aware of the community's concerns around the future of things like Widget, UiBinder, etc. and how GWT 3 might upend the apple-cart.  GWT 3 is still very far away and its plans very nebulous.  It's far too early to talk specifics, but I assure you that we are keeping a close eye on it and planning for that future. Whatever the outcome, it is likely that you will find a way forward with Sencha.

That said, a community project to pick up the mantle for Widget, UiBinder and other pieces that get dropped from GWT is an excellent idea.

[Thomas Broyer]

No need to create a new project; that can live in GWT proper BUT will only be supported by community members (i.e. not Google)

[Goktug Gokdogan]

I think creating a new project is a good opportunity to start bundling things out of GWT-SDK as agreed on in the initial roadmap.

On Thu, Apr 28, 2016 at 12:27 AM, Thomas Broyer <t.br...@gmail.com> wrote:

No need to create a new project; that can live in GWT proper BUT will only be supported by community members (i.e. not Google)

--
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 google-web-toolkit-co...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/google-web-toolkit-contributors/ad848b18-5451-4b7f-9c8f-bb58b86a9b81%40googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Thomas Broyer]

Sorry, I didn't mean to keep a big "monolithic" project like we have now (I'm +1000 on modularization!), it can very well be a new "project", built separately, released at a different pace, etc.

My point was that it could be kept under the "GWT Project" umbrella, rather than a "third party" project. It can also be incubated as a third-party *now*, as we don't yet really know what GWT 3 will be, and then brought under the GWT Project umbrella when we know what GWT 3 will and won't be.

For such things like widgets and UiBinder, I think the "GWT Project" should provide a migration path from GWT 2 to GWT 3.

On Thursday, April 28, 2016 at 9:36:34 AM UTC+2, Goktug Gokdogan wrote:

I think creating a new project is a good opportunity to start bundling things out of GWT-SDK as agreed on in the initial roadmap.

On Thu, Apr 28, 2016 at 12:27 AM, Thomas Broyer <t.br...@gmail.com> wrote:

No need to create a new project; that can live in GWT proper BUT will only be supported by community members (i.e. not Google)

--
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 google-web-toolkit-contributors+unsubscribe@googlegroups.com.

[Paul Stockley]

I think once 2.8 is out the door we should definitely pursue setting up some projects to tackle the things that won't be in J2CL. This will really help a lot to alleviate many peoples fears, especially those with large projects that use Widgets, UiBinder etc. 

There has been a lot of vague talk about the future up to this point and to be honest, talk is cheap. We need some real action. I think if there are smaller independent projects that are maintained in Github and don't need to use Gerrit, you will get way more people contributing. The good thing is that these new projects could potentially be used with 2.8 and allow people to start the migration process sooner rather than later. 

I guess the first steps would be to define what needs to be extracted from GWT core and define the goals/non-goals for each project. Our company is definitely interested in:

Basic Widget support for evergreen browsers and IE 11 (we only use Widget and a few of the base panel classes)

Resource  bundles/ GSS

UiBinder.

We will be wiling to help make these projects a reality.