On Wednesday 10 March 2004 13:14, Daniel Chudnov wrote: > > As for the get-submission problem, having seen it come up a few times > I'm confused about the particulars of this. One of my forms is pre-fed > by a GET url like "/path/to/form?p1=foo&p2=bar". The form also has a > submit button, added like this: > > form.add_submit('spam', 'spam') > > Instead of just calling the canonical "if not form.is_submitted(): > return render()", there are instead two possible calls to render(): > > if not form.is_submitted(): > return render() > > if form.get_submit() != str('spam'): > return render() > > ...so GET calls with the params listed above still drop back to render() > when is_submitted==True. > > Am I missing a bigger-picture issue? Perhaps this feels too much like a > hack to others? Because to me it seemed like elegant, not insufficient, > design. :) > Daniel, I think your example is fine. I just wanted to point out what I think, from the list messages, is a rather general case that is GET method requests being used, as your case, for pre-fedding form inputs. And also wanted to know other opinions about if it's so general or exclusive. In my case, I only use GET method requests for submitting forms in a vey few cases, and so for my convenience I use the following condition in my modified is_submitted method in the Form subclass. bool(request.form) and ( request.get_method() == "POST" or DEFAULT_SUBMIT ) where DEFAULT_SUBMIT is a class attribute that use to be False. As Jason points out , widgets still need to be subclassed. On Wednesday 10 March 2004 13:47, Neil Schemenauer wrote: > > That's all I can think of right now. If you don't mess with > Component objects then perhaps the visible changes will not be > extensive. Hopefully. :-) > Possibly, widgets wouldn't need to be wrapped inside components, but I like the actual three levels of flexibility: add(), add_component(), _add_name() then might be component class could act as a kind of widget prototype. I look forward to see these changes. Oscar Rambla