[Zope3-dev] a plan for widgets?

Martijn Faassen faassen at infrae.com
Fri Mar 17 04:08:18 EST 2006


Gary Poster wrote:
> 
> On Mar 16, 2006, at 11:42 AM, Martijn Faassen wrote:
> 
[snip]
>> Roger Ineichen wrote:
>> [snip]
>>
>>> btw,
>>> didn't Gary Poster start a widget refactoring?
>>
>>
>> Good question. I have no idea what the status of all that stuff is;  
>> whether anything got merged or not..
>  
> At ZC, several of us had enough experience with the zope widgets that  
> we felt pretty confident that we wanted to start again, with a  
> backwards compatibility story.  We started, but ran out of steam, and  
> didn't have a driving project for the changes, so we shelved it.
> 
> We have an upcoming project that will want the changes.  Our current  
> plan is to develop what we need as "zc.widget" or something, and open- 
> source it at the end when it's what we need, in the hopes that some  
> will find it compelling enough to join in the maintenance and further  
> development (btw, thanks, dobe, for the work on resourcelibrary!).   No 
> public timeframe.

What's 'dobe' mean?

> Doing it in the "zc" namespace is intended to make no assumptions  about 
> what happens with zope.app.form; zc.widget probably will have  some 
> backwards-compatibility with zope.app.form, but no dependencies.

Hm, while redoing widgets to be better and presumably more capable is 
definitely cool and interesting, some things have got me a bit worried, 
in particular the "no public timeframe" and the need for my projects to 
work with widgets.

I have some questions about the new design - how fundamental is this 
change to be? Will the widgets API change drastically and will this 
require changes in formlib as well? Is this part of the backwards 
compatibility story? It doesn't sound right if it is, as there would be 
no point to a new API if the forms machinery would use the old one.

What also got me worried is that the promise of future new widgets may 
result in the community doing very little for the time being. We seemed 
to have a little bit of momentum to clean up zope.app.form.browser, but 
this will now evaporate most likely. It also potentially makes efforts 
like using Zope 3 widgets with Zope 2 archetypes not very worthwhile 
right now, for instance.

On the other hand, I can see why, if you have clear ideas about a 
design, you'd want to develop this in isolation and move it to open 
source later. But it still isn't positive for any community-driven actions.

What to do about this?

Regards,

Martijn


More information about the Zope3-dev mailing list