Hi,
we have a rails 1.0 application and I did some tests to see if we could
upgrade
to rails 1.1
One issue I found, is that prototype 1.5.0_rc0, that was installed by
rake
rails:update:javascripts, changes the way a select list is serialized if
an
option has an empty value.
Consider a form like
<%= form_tag :action => ‘test_form_post’ %>
<% options = { ‘’ => ‘select’, ‘1’ => ‘one’, ‘2’ => ‘two’ } %>
<%= select_tag(“test”, options.keys.collect { | key | “#{options[key]}” }.join() ) %> <%= submit_tag
“send” %>
<%= end_form_tag %>
which is rendered to
select one two as expected.The idea is to present the entry ‘select’ initially to indicate that the
user
should select an option. In case he does not, an empty value is
returned, which
fails in a validates_presence_of validation.
This works for normal forms and for remote forms using rails 1.0 /
prototype
1.4. In both cases I get a parameter hash
{“commit”=>“send”, “action”=>“test_form_post”, “controller”=>“develop”,
“test”=>""}
Unfortunately rails 1.1 / prototype 1.5.0_rc0 provides
{“commit”=>“send”, “action”=>“test_form_post”, “controller”=>“develop”,
“test”=>“select”} for a remote form.
Instead of the empty value the content of the option is returned.
I think this would be ok, if there wasn’t a value attribute at all (in
that case
form submission would output the content as well).
Is this difference between normal form submission and remote forms
intentional?
It was one of the great features that prototype allows me to switch from
forms
to remote forms without having to change anything in the handling of the
form
data.
If it is intentional: how should my use case be implemented?
I want to have a labelled option entry, that indicates, that the user
should
make a choice (‘select’ is just the label I choose for the test case).
But I don’t want to deal with that label in the application.
Of course I could use some illegal value like -1 in the app and do some
special
treatment. But that implies extra code in the contoller or model.
regards
Morus