myfaces-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Werner Punz <werner.p...@gmail.com>
Subject Re: Dojo discussion - opensourcing the jsf dojo components project
Date Mon, 14 Jul 2008 07:59:36 GMT
Andrew Robinson schrieb:
> I can see the point of that argument, but worry that putting heavy 3rd
> party JS libraries into Tomahawk will steer people away from using it.
> IMO, Dojo based components should either be (1) in a new MyFaces top
> project, or in (2) a subproject of Tomahawk (i.e.
> myfaces-tomahawk-dojo).
> 
Sorry this is getting long.

Actually the whole point for starting the work (and believe me it is a 
load of work especially since I cannot work fulltime on it)
was to start a project which in the long run should eliminate the
dojo dependency out of tomahawk, that goal still is valid.

My entire idea was to make maybe an extension project
tomahawk-extensions or something alike
which should allow to integrate third party dhtml libs.
One aim is that the existing components should work with the new ones.
Which is an easy aim since tomahawk is loosely coupled so are the new 
components.
The second aim was, if we integrated dojo, this time do it right by 
doing a thin jsf layer covering most components,
and keeping those up to date with as much code generation as possible.
(This goal is mostly fulfilled although I am relying currently
on my specialized codegens utilizing groovy for component descriptions)

Third aim was to get the dojo sources out of the sourcetree,
which is possible by now as well, by relying on weblets!


The other thing is since I relied on it being a tomahawk extension, I 
took the liberty and added tomahawk dependencies wherever it made sense
(shared stuff etc...)

Anyway to get out of the technical stuff again:

I am fine with either approach if the decision is more along the lines 
of having it in the sandbox I am fine with it as well although I am 
uneasy with it at the current stage of the project.

I however feel somewhat uneasy to add all this to the core tomahawk 
stuff, I personally thing that core Tomahawk, and that is an opinion 
which is different from two years ago.

Should not rely too much on third party libs (hence starting the entire 
dojo project),
and should be as solid as possible.

Third party libs etc.. which are worth integration should get into
their own extensions.
This would make it clear that this is Tomahawk, and you can add 
extensions which should work fine with it if you need additional 
functionality.

Anyway this does not really help me for now how to host the code in 
development. A code grant is fine, I will take care of all this this week.
But I want to move the code over slowly to opensource, which means some 
public repo somewhere without users starting to adopt the stuff for now
(unless they really know what they do).
The code grant should be no problem every line of code for now is 
written by me and developed mostly under the irian umbrella in my 
opensource donation time. So getting all who have to sign to sign it is 
no problem.


But:
Would it be possible to keep the code on the branch for a while and once 
an integration stage is reached to move it over to the trunk?

I dont think I am ready yet to go into the sandbox, due to the fact that 
as soon as it hits the sandbox we have at least 100 users and 2 tool 
vendors starting the integration! While many of the components are 
usable if you know what to use from the properties it is not sandbox 
ready yet! Due to the lack of documentation (which also will be 
generated to a big degree) and a lack of cleanup on the property list of 
every component, also to the fact that I have not started yet to move 
the sandbox components over!



Werner


Mime
View raw message