Dojo plugin update using 1.1.0 framework

[Jeromy Evans - Blue Sky Minds]

Dave Newton wrote:
>
> Nutshell: what's anybody's take on the effort this would require, and who's
> available to make that effort?
>
>  

I share similar sentiment and at most will just be able to convert my
existing 2.1.1 test applications over to use the Dojo 1.x plugin to
investigate the consequences.

I've concluded that tag libraries for rich client framework like these
no longer warrant the effort.  I prefer to encourage users to use the
client-side libraries as intended by their designers, whether that's
with HTML markup or programmatic javascript. This approach benefits the
users as they can receive support directly from Dojo or other vendor,
can develop skills that can be transferred between server-side
frameworks and they have no artificial constraints placed on them by a
tag library. Struts2's role should be to simplify server-side
integration, such as by providing default models that can be used by the
Dojo widgets.  And provide really good examples.

On a related issue, improving support for JSF components may allow user
to leverage existing tag libraries instead of re-inventing them.


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@...
For additional commands, e-mail: dev-help@...

[Martin Cooper-2]

On Sat, Apr 5, 2008 at 5:35 PM, Jeromy Evans <
jeromy.evans@...> wrote:

> Agreed.  The Dojo 0.43 plugin in Struts2.1.1 contains significant
> improvements over the Dojo 0.40 tags bundled in 2.0.x.  It's worth releasing
> as-is and I'd give it a +1 today.
>
> It sounds like there's enough people interested to complete a Dojo 1.x
> plugin.  I also think it's worth creating a googlecode project for it until
> it reaches a certain level of maturity.  The benefit of googlecode over the
> sandpit is the low barrier to contribute and informal releases.


If you start the project at Google Code, remember that you will have to come
through the Incubator, one way or another, to get the code back into Struts,
even if that's just the IP Clearance route. And if, by "low barrier to
contribute", you mean that you want to grant committership to people who are
not Struts committers, then remember that either those people will lose
commit rights when you bring the code back here, or you will have to go
through a different incubation process to deal with the additional
committers.

On the other hand, creating the project in the Struts sandbox means that it
is immediately open to any Struts committer, all of the resources are
already set up, and getting a release out is dependent only upon a vote to
move the code from the sandbox to the main code line. I'd say that path will
be a whole lot less hassle - unless, that is, you expect the Dojo 1.x plugin
to be a major project that requires additional committers and spans an
extended period of time to get into shape equivalent to that of today's Dojo
plugin.

--
Martin Cooper

>
> Musachy Barroso wrote:
>
> > I don't think we should wait at all. Refactoring dojo out of core was
> > one of the main things for 2.1 and it's been there for a year already.
> > Unless Dojo 1.0 is a lot, way, way better than the older versions, I
> > would say you will find lots of surprises. IMO you should set it up as
> > a project on googlecode or somewhere else and we can all
> > contribute/test and eventually bring it on (or just keep it there and
> > get rid of our current plugin).
> >
> > my 2 centavos :)
> >
> > musachy
> >
> > On Sat, Apr 5, 2008 at 6:19 PM, Dave Newton <newton.dave@...>
> > wrote:
> >
> >
> > > --- Al Sutton <al.sutton@...> wrote:
> > >  > Whilst I can see that there is an advantage to getting a 2.1
> > > release out,
> > >  > my question would be do we want it to go out with a (very) old
> > > version
> > >  > of dojo as the demonstration of it's modern ajax capabilities?, and
> > > do
> > >  > we want to put developers through getting used to 0.4 as the
> > > bundled
> > >  > version and then jump to a much newer version as a minor version
> > > release?
> > >
> > >  So the questions are how well tested are the new Dojo tags, and if
> > > they're
> > >  not tested well enough how long would it take to test them? Lastly,
> > > how much
> > >  rework, if any, is required to match the functionality of the 0.4x
> > > plugin?
> > >
> > >  My impression is still that Dojo 1.0 is pretty different from Dojo
> > > 0.4x, and
> > >  that this is a non-trivial project--but that's a guess made from
> > > ignorance.
> > >  Is there any evidence to the contrary? Have the tags been tested
> > > (even
> > >  manually) on the client side to bulk-verify behavior?
> > >
> > >  Due to some immediate responsibilities, my availability for working
> > > on a Dojo
> > >  1.0 plugin is limited and conditional:
> > >
> > >  -- I have some time I can dedicate to *testing* new Dojo tags.
> > >  -- I don't have the time to learn Dojo 1.0 well and implement much
> > > changed
> > >  and/or new functionality if both the cost and risk are high.
> > >  -- The window within that time is available is short, and dwindling.
> > >  -- The more people working on it the more likely I am to make the
> > > time
> > >  because of a perceived lower risk.
> > >
> > >  Nutshell: what's anybody's take on the effort this would require, and
> > > who's
> > >  available to make that effort?
> > >
> > >  Dave
> > >
> > >
> > >
> > >
> > >  ---------------------------------------------------------------------
> > >  To unsubscribe, e-mail: dev-unsubscribe@...
> > >  For additional commands, e-mail: dev-help@...
> > >
> > >
> > >
> > >
> >
> >
> >
> >
> >
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@...
> For additional commands, e-mail: dev-help@...
>
>

[dkarr]

> -----Original Message-----
> From: Jeromy Evans [mailto:jeromy.evans@...]
> Sent: Saturday, April 05, 2008 5:56 PM
> To: Struts Developers List
> Subject: Re: Dojo plugin update using 1.1.0 framework
>
>
> I've concluded that tag libraries for rich client framework
> like these no longer warrant the effort.  I prefer to
> encourage users to use the client-side libraries as intended
> by their designers, whether that's with HTML markup or
> programmatic javascript. This approach benefits the users as
> they can receive support directly from Dojo or other vendor,
> can develop skills that can be transferred between
> server-side frameworks and they have no artificial
> constraints placed on them by a tag library. Struts2's role
> should be to simplify server-side integration, such as by
> providing default models that can be used by the Dojo
> widgets.  And provide really good examples.

I also feel strongly that this direction holds the best long-term value.
However, I don't believe that Dojo and YUI and others like it should
just be ignored by Struts2 and others like it.  In fact, the server-side
frameworks should embrace and facilitate the combination, either with
light integration interfaces, or simply with good documentation and
examples that describes how to combine them effectively.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@...
For additional commands, e-mail: dev-help@...

[Al Sutton]

Given the discussion, how about the following idea;

- Remove the dojo plug-in from the S2 codebase and put it, in it's current
state, into a googlecode project.

My main driver for doing the updates was to update the S2.1 code to bring it
inline with the latest version, but perhaps we should put our hands up and
admit that S2's focus is not on Ajax UI widgets and recommend that web
developers go out and use whichever framework they want as opposed to
limiting them to the choice made by the tag developers. After all, there is
little in the way of S2.1 code for ajax, all we have is some code that acts
as a wrapper to make dojo look like it's part of S2.

We can put in the readme a pointer to the googlecode project, and see how it
develops from there.

What do people think?

Al.

----- Original Message -----
From: "Martin Cooper" <martinc@...>
To: "Struts Developers List" <dev@...>
Sent: Sunday, April 06, 2008 3:16 AM
Subject: Re: Dojo plugin update using 1.1.0 framework

> On Sat, Apr 5, 2008 at 5:35 PM, Jeromy Evans <
> jeromy.evans@...> wrote:
>
>> Agreed.  The Dojo 0.43 plugin in Struts2.1.1 contains significant
>> improvements over the Dojo 0.40 tags bundled in 2.0.x.  It's worth
>> releasing
>> as-is and I'd give it a +1 today.
>>
>> It sounds like there's enough people interested to complete a Dojo 1.x
>> plugin.  I also think it's worth creating a googlecode project for it
>> until
>> it reaches a certain level of maturity.  The benefit of googlecode over
>> the
>> sandpit is the low barrier to contribute and informal releases.
>
>
> If you start the project at Google Code, remember that you will have to
> come
> through the Incubator, one way or another, to get the code back into
> Struts,
> even if that's just the IP Clearance route. And if, by "low barrier to
> contribute", you mean that you want to grant committership to people who
> are
> not Struts committers, then remember that either those people will lose
> commit rights when you bring the code back here, or you will have to go
> through a different incubation process to deal with the additional
> committers.
>
> On the other hand, creating the project in the Struts sandbox means that
> it
> is immediately open to any Struts committer, all of the resources are
> already set up, and getting a release out is dependent only upon a vote to
> move the code from the sandbox to the main code line. I'd say that path
> will
> be a whole lot less hassle - unless, that is, you expect the Dojo 1.x
> plugin
> to be a major project that requires additional committers and spans an
> extended period of time to get into shape equivalent to that of today's
> Dojo
> plugin.
>
> --
> Martin Cooper
>
>
>
>>
>> Musachy Barroso wrote:
>>
>> > I don't think we should wait at all. Refactoring dojo out of core was
>> > one of the main things for 2.1 and it's been there for a year already.
>> > Unless Dojo 1.0 is a lot, way, way better than the older versions, I
>> > would say you will find lots of surprises. IMO you should set it up as
>> > a project on googlecode or somewhere else and we can all
>> > contribute/test and eventually bring it on (or just keep it there and
>> > get rid of our current plugin).
>> >
>> > my 2 centavos :)
>> >
>> > musachy
>> >
>> > On Sat, Apr 5, 2008 at 6:19 PM, Dave Newton <newton.dave@...>
>> > wrote:
>> >
>> >
>> > > --- Al Sutton <al.sutton@...> wrote:
>> > >  > Whilst I can see that there is an advantage to getting a 2.1
>> > > release out,
>> > >  > my question would be do we want it to go out with a (very) old
>> > > version
>> > >  > of dojo as the demonstration of it's modern ajax capabilities?,
>> > > and
>> > > do
>> > >  > we want to put developers through getting used to 0.4 as the
>> > > bundled
>> > >  > version and then jump to a much newer version as a minor version
>> > > release?
>> > >
>> > >  So the questions are how well tested are the new Dojo tags, and if
>> > > they're
>> > >  not tested well enough how long would it take to test them? Lastly,
>> > > how much
>> > >  rework, if any, is required to match the functionality of the 0.4x
>> > > plugin?
>> > >
>> > >  My impression is still that Dojo 1.0 is pretty different from Dojo
>> > > 0.4x, and
>> > >  that this is a non-trivial project--but that's a guess made from
>> > > ignorance.
>> > >  Is there any evidence to the contrary? Have the tags been tested
>> > > (even
>> > >  manually) on the client side to bulk-verify behavior?
>> > >
>> > >  Due to some immediate responsibilities, my availability for working
>> > > on a Dojo
>> > >  1.0 plugin is limited and conditional:
>> > >
>> > >  -- I have some time I can dedicate to *testing* new Dojo tags.
>> > >  -- I don't have the time to learn Dojo 1.0 well and implement much
>> > > changed
>> > >  and/or new functionality if both the cost and risk are high.
>> > >  -- The window within that time is available is short, and dwindling.
>> > >  -- The more people working on it the more likely I am to make the
>> > > time
>> > >  because of a perceived lower risk.
>> > >
>> > >  Nutshell: what's anybody's take on the effort this would require,
>> > > and
>> > > who's
>> > >  available to make that effort?
>> > >
>> > >  Dave
>> > >
>> > >
>> > >
>> > >
>> >
>> > >   ---------------------------------------------------------------------
>> > >  To unsubscribe, e-mail: dev-unsubscribe@...
>> > >  For additional commands, e-mail: dev-help@...
>> > >
>> > >
>> > >
>> > >
>> >
>> >
>> >
>> >
>> >
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: dev-unsubscribe@...
>> For additional commands, e-mail: dev-help@...
>>
>>
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@...
For additional commands, e-mail: dev-help@...

[Jeromy Evans - Blue Sky Minds]

Martin Cooper wrote:

>
>  
> On the other hand, creating the project in the Struts sandbox means that it
> is immediately open to any Struts committer, all of the resources are
> already set up, and getting a release out is dependent only upon a vote to
> move the code from the sandbox to the main code line. I'd say that path will
> be a whole lot less hassle - unless, that is, you expect the Dojo 1.x plugin
> to be a major project that requires additional committers and spans an
> extended period of time to get into shape equivalent to that of today's Dojo
> plugin.
>
> --
> Martin Cooper
>
>
>
>  

I agree transition from googlecode back into Apache is an important
consideration and a googlecode decision should be made only after we
understand the level of effort involved.

My gut feeling is that a complete Dojo 1.x plugin requires significant
effort and additional committers. Others would know better than me
though.  Do you have a contact over at the Dojo that :
  - is familiar with the current Struts Dojo capabilities and can give
an indication of the effort to migrate to 1.x; or
  - more importantly, confirm whether recreating custom tags is a
sensible approach at all.

It may be better for both projects that we make a concerted effort to
demonstrate dojo + struts integration rather than creating another
plugin that encapsulates Dojo.


---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@...
For additional commands, e-mail: dev-help@...

[newton.dave]

--- Al Sutton <al.sutton@...> wrote:
> My main driver for doing the updates was to update the S2.1 code to bring
> it inline with the latest version, but perhaps we should put our hands up
and
> admit that S2's focus is not on Ajax UI widgets

Nobody has yet provided any information as to the potential cost of making
Dojo 1.0 work and nobody answered my question regarding what level of
functionality currently exists in the Dojo 1.0 plugin, so I'm pretty much
unable to come to any cogent conclusion.

If it came to a vote I'd probably say (a) leave the current Dojo plugin where
it is, and (b) put any new tag plugins on Google Code (like my semi-dormant
jQuery plugin and others people are working on).

I don't know what the ASF policy is on "endorsing" particular code, but it
seems like if a new plugin got to the point of being usable it would be
simple enough to add a "we currently recommend using..." or "among the better
tag libraries are..." or whatever.

Dave



---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@...
For additional commands, e-mail: dev-help@...

[Al Sutton]

Dave,

I don't think anyone does have a good handle on how much work it would be to
integrate Dojo 1.0 into the S2.1 tag framework. It's one of those things
that the time taken to properly evaluate and estimate the work necessary may
be longer than the time to do the work, and even when the work is under way
there may be some unforeseen issues even if an estimate was made. The Dojo
guys have tried to make it easier by providing an api change list from 0.4
to 0.9 then from 0.9 to 1.0, but how that applies to the S2.1 code is
something that will only make sense to someone who has the code fresh in
their mind.

This is why I started to work on the changes and kept the code as a separate
entity, that way if it does prove to be a mountain rather than a molehill
there's nothing being held up, and if it turns out to be a molehill then we
get all the benefits of supporting, matured code with a smaller footprint in
a short period of time.

Al.

----- Original Message -----
From: "Dave Newton" <newton.dave@...>
To: "Struts Developers List" <dev@...>
Sent: Sunday, April 06, 2008 3:18 PM
Subject: Re: Dojo plugin update using 1.1.0 framework

> --- Al Sutton <al.sutton@...> wrote:
>> My main driver for doing the updates was to update the S2.1 code to bring
>> it inline with the latest version, but perhaps we should put our hands up
> and
>> admit that S2's focus is not on Ajax UI widgets
>
> Nobody has yet provided any information as to the potential cost of making
> Dojo 1.0 work and nobody answered my question regarding what level of
> functionality currently exists in the Dojo 1.0 plugin, so I'm pretty much
> unable to come to any cogent conclusion.
>
> If it came to a vote I'd probably say (a) leave the current Dojo plugin
> where
> it is, and (b) put any new tag plugins on Google Code (like my
> semi-dormant
> jQuery plugin and others people are working on).
>
> I don't know what the ASF policy is on "endorsing" particular code, but it
> seems like if a new plugin got to the point of being usable it would be
> simple enough to add a "we currently recommend using..." or "among the
> better
> tag libraries are..." or whatever.
>
> Dave
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: dev-unsubscribe@...
> For additional commands, e-mail: dev-help@...
>

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@...
For additional commands, e-mail: dev-help@...