Still there!

[temsa]

Hi, all!

It's been a while I haven't written anything here. And I'm really
sorry for that.
Archetype isn't dead, but a year ago both Swiip and I did have some
life changing situations, so working on was way more complicated.
I have continued to develop Archetype, and even made a 0.10.1 to fix
problems with firebug logger. However I didn't make the zip/tar.gz, so
it's only available as a tag in the main repository yet.

I have used a 0.11 one on a big project @work and was quite happy with
it.
I made an url managing library which is very useful for mapping an url
to an application state, so it's easy to have both history, and real
bookmarkable links in your ajax application (using the url Target for
virtual paths)
I have practically entirely rewritten zed archetype compiler as a java
maven plugin, but did not finished the work yet.

I tried to use some other libraries/frameworks to see if archetype is
still useful or not. In fact, it is.

I would like archetype to go the commonjs way, and right now the
discussion about a unified module 2.0 specification is very important
to me. I would like archetype to be easier to use and understand.

I would like archetype to be quite different in it's use, but with the
same objective in mind, and I don't know if I should keep the
archetype name.
Archetype right now can jump in any website, but it's particularly
suitable for RIAs, so I would like to focus only on this: having a
good default template system, having an easy way for component
cascading, make a smart use of html5 features and bring them to old
browsers (or not ?).
In order to keep it simple I'd like to do some plugin system, maybe
the jquery way.

This sounds as a total rewrite, and it will take time, or many
developers and contributors. Are there any people here who would like
to jump on the boat and become an official developer/contributor?

Cheers,
florian

[Matthieu Lux]

Hi,

Still there too! As Florian said, life situations changing and, we must admit, no real (public) community or activity on our project killed my implication but I’m still there.

If Florian got chances to try other JS solutions, my recent experiences bring me to Flex. I think this experience is very interesting because Flex got exactly same problematic than classic Web RIA and got many different answers.

Flex is really far than the perfect solution. I think about compilations problems, weight of the framework, quality of the components… even if Flex 4 bring some good. But there is some really good points: data binding is an incredibly simple but efficient idea (I read that recent version of JQuery, I must dig that), an important library of UI components is really useful and Granite DS is far more powerful than any JSon framework I ever seen.

To come back on Archetype, I will not repeat all Florian already said but just keep important points: yes, Archetype is still useful, yes it is too hard to jump in, yes, I think that the name Archetype cannot be kept and yes, a full rewrite is unavoidable.

If we want to move on, my main idea will be to work first on concepts and notations we want: finding standards and diagrams to build from. I think about (Yes I’m a Spring fan) Spring JS why not, or also MVC standards like Pure MVC.

My second idea is to break some rules we kept on Archetype: address problem of the server communication (I must have what Granite DS do), build what we need to having a real project management with JS Maven artifacts, not allow every time multiple implementations of everything (It’s already complicated enough), address some real implemented components and finally but it is certainly the most difficult: not binding ourselves to another Framework (in fact Archetypes used really not many functionalities of Prototype).

Last idea I want to insert: we also must have a demonstration project to experiment the framework.

Oh dear, I written a mail bigger than Flo!  I will stop here; I think I launch enough trolls for one mail, I’m curious to know Florian’s position on all this points. I hope we’ll be able to draw the main lines of a new Archetype project?

Matthieu

PS: Sorry for my English, no English corrector other than Word available this time ;)

2010/10/30 temsa <florian....@gmail.com>

--
You received this message because you are subscribed to the Google Groups "Archetype Javascript Framework - User Group" group.
To post to this group, send email to archety...@googlegroups.com.
To unsubscribe from this group, send email to archetype-use...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/archetype-user?hl=en.

[Florian Traverse]

Hi Matthieu !

Glad to see you answering Matthieu !!!!! :) :) :) :)

During this year, I had to do some Flex/AS3 too. Certainly far far less than you, but still, I did some. And I agree there are some good things in there.

Now with :

- web browser acceleration

- JS getting 4-7x faster than at the start of the year

- JS 10 times faster than the end 2008

- good CSS2 / 3 support (if you want, I can give you some life changing, logical, css recipes for modern browsers + shims for IE<8 for your layouts, thaht could easily be enhanced for mobile devices with smaller screen)

- good SVG support including in IE9

We can do a lot more in the browser at the potential cost of telling people "upgrade your browser or get google chrome frame plugin", so Flash/Flex for me is not necessary for a lot of things anymore. Even Microsoft is going to HTML5/JS (I hope this is not an "embrace, extend&extinguish" strategy), rather than WPF or Silverlight for a lot of their future API, including on the desktop (yes it has been an internal war @Microsoft that gived this result less than a week ago).

I'm totally ok to build on standards/reference libraries/frameworks like what we can find in Spring or Pure MVC, and I try to see as large as possible for a source of inspiration

I agree, we shouldn't depend on another frameworklike what we did with Prototype/jQuery, but just play nice if such a framework is here. Almost any Browser JS development I know yet uses jQuery right now, so we at least have to play nice with it, and focus on doing other thing. I'm pretty convinced however we can make useful thing for the middle/long term, by explaining our needs to the CommonJS community : they are pretty smart, a bit too server centric, but they are currently working on a new module spec that really cares about the current browsers.

Building on CommonJS, and CommonJS Commons would provide a way to have JavaScript modules that would run both on the server side or the browser side, which is more than smart (imagine running Archetype or parts of it, serverside, at about no cost!) :). CommonJS provides a package system that could be seen as pretty similar as as a maven pom in some ways, so maybe a CommonJS package plugin for Maven would do the trick, while not enforcing Maven or Java use.

I'd like not to depend on a framework, but I've looked at some cool libraries, and we can use/integrate or even fork some of them like RequireJS, FlyWrap or Yabble to manage for exemple "module loading". FlyWrap is nice and quite simple and I think I will experiment some propositions for Module 2.0 spec by forking it from github. RequireJS is the best module loader right now for Module 1.0/1.1 in the browser, it manages the dev mode/production mode I love, and plays nice with jQuery, and can be used as a loader for Dojo if I remember.

On the feature level, I still have to play more with SproutCore, which is not bad at all, but needs Ruby to make a server for development, and uses "compilation" in dev mode(even if the server takes care of doing it automagically), and I don't like this. By the way, it is well thinked, can be a good inspiration source, because it is a kind of port of Apple's SDK to make Desktop Application to JS/Browser, so the principles are pretty good (e.g. the fixtures : basically sample data for running the website without a real server, or another example, it has the nice value binding feature too );) even if I don't like everything in it.

For the demonstration, we can use the booking demo, as it allows for comparison between frameworks pretty easily. I have no other idea yet, but any is welcome.

As a conclusion, I'm very exited to work on such a project again, and I would also like to make the project in a Bazaar mode, not in Cathedral mode, this time, please people join and help us :)

Cheers,

Florian

2010/11/3 Matthieu Lux <matthi...@gmail.com>

[Matthieu Lux]

Hi,

You don’t need to convince me that Web standards in a good browser are better than launching a Flash player. I think we shouldn’t start this troll here; we’ve already enough subjects to treat.

In order to start working on the design of “the” project, I think we must focus right now on the way we will be working. I have some ideas on this subject:

•           The most important point for me is to sequence the questions we have to discuss. I think we should start to agree on the principal goals (not 15, 5 at max), perhaps find a new name but not digress on more specific questions before agreed on them. We both have passion for Web and JS and we both can digress enough to never take decisions and move forward.
•           I’m not sure that the “Archetype User Group” is the best place for that, perhaps open a new one (place)? But is it interesting to lose all people having subscribed to this one? I don’t know…
•           I’m not sure the mail is the better way either. (Ok I never liked mails) What about blog or twitter? Most of the times, blog posts are more readable because we better organize a main idea, and we have the comments solution to react.

I hope this method can bring more interested people in order to start a community. And to answer to the end of your mail, I think that the question of involvement in CommonJS and for what is exactly a subject to be discuss after knowing what we want to do.

Matthieu

2010/11/3 Florian Traverse <florian....@gmail.com>