> On 7/23/07, Michael Koziarski <mich...@koziarski.com> wrote:

>> Without cases which
>> can't be solved by this pattern, I'm not sure that adding a new
>> callback is justified.

> Sorry for replying to myself, but the wireless here is a little too
> spotty to reply to all the individual threads.

> Providing a method to allow static defaults such as:

> set_default_values :foo=>'bar'

> Is just reimplementing the pre-existing defaults code that we extract
> from the columns, so I'm not sure that's the right way to go.

> For dynamic defaults, such as 'the default value for the categories on
> a blog, is taken from the account's default values' I've not seen a
> case that can't be catered for by overriding the initialize method.

> Requiring users to call super first is perfectly acceptable to me.  To
> expect to use methods from a super class without first initializing
> that super class seems very strange to me.

> Of course, there may well be particular cases where it's genuinely
> confusing that something doesn't work, or that the error messages are
> misleading,  but we can fix those on a case by case basis.

> Every feature that we add, becomes something we need to support on an
> ongoing basis, for the foreseeable future.  So we have to have a
> reject by default, or we'll end up with another 'components'.

> Koz