dotLRN's supposed to deploy in a couple of weeks and OF and furfly are working actively on the OpenACS tip so if your code were to break things a few of us might be unhappy. OF uses the form builder heavily (yay!) and furfly has too in our two packages.
How well has this been tested?
I've got some trivial commits of my own to make to acs-templating, including making "date" a real type rather than simply a funky widget (it uses the same widget so everything's backwards compatible). I need these for my (modest) dotLRN work.
I like the "-mode display" addition. ad_form currently supports "-confirm_template" which allows you to specify a template in the "OK Cancel" style displayed for confirmation before ad_form executes the "-new_data" or "-edit_data" clauses. Your "-mode display" gives ad_form a way to generate confirm pages without the need for the user to write their own confirmation template, and that's cool. A user will then be able to use "-confirm" if they're lazy/satisfied with the automatically generated page, "-confirm_template" if they want more control. Yeah, sounds pretty cool alright!
"-buttons" is good, too. I did something similar for Greenpeace but made a "button" datatype with widget instead of adding a special switch. We might want to compare notes before deciding which approach is best. Something that allows us to generate multiple buttons is certainly needed.
The only advantage of having it as a type is that the user can insert it anywhere in the form, and can use it with standard element stuff like "html" (think JavaScript).
But my Greenpeace code for this is a bit of a hack, IIRC ...
Maybe you and I can test this a bit more, decide which "button" approach is best, and I can add some integration with ad_form and then we can commit it? In (say) the next week or so?
