Form generation libraries that contain validation are very tough to generalize 
in a way that lends itself to framework convenience.  I think Formish does a 
pretty good job here, because it physically separates out a lot of stuff that's 
all glommed together in other system (and thus there's schemish, validatish, 
convertish, and even 'dottedish').  It's shows maturity in design.

But at its very outer layer it *does* depend on the request having a particular 
API (the WebOb API).  But so does BFG, so this works out pretty well.

I suspect it will be a bit of a project to use anything that depends on a 
particular request API but doesn't depend on the WebOb API.

So I have the beginnings of BFG-ish bindings for Formish here (unfinished and 
very unstable; the docs are currently total bollocks and don't match the 

I hope to have a release of this out within the next few weeks.

Martin Aspeli wrote:
> Chris McDonough wrote:
>> +1 to what Martin wrote, with something else that can only add to the 
>> confusion:  recently I have been using Formish ( to do 
>> autogenerated forms.  I think both zope.formlib and z3c.form try to paint 
>> forms 
>> based on "model" objects; formish doesn't even try.  It just lets you create 
>> your own "form schema" and attempts to draw forms based on that.  This is 
>> enough for my needs currently.
>  From a quick read of the documentation, formish looks nice. It also 
> looks very similar to z3c.form, and has many of the same abstractions 
> (fields, widgets, form renderers, data converters).
> z3c.form indeed contains some base classes for "model-based" 
> add/edit/display forms, but it's easy to base it on anything (you don't 
> even need an interface, you can just pass it a sequence of schema fields).
> In this regard, z3c.form is at least a lot more flexible than zope.formlib.
> Martin
> _______________________________________________
> Repoze-dev mailing list

Repoze-dev mailing list

Reply via email to