more work on idp-distribution

[Cantor, Scott]

On 8/23/13 4:37 PM, "Marvin S. Addison" <marvin.addison at gmail.com> wrote:

>The declarative view handling is simple in terms of usage. In terms of
>performance I suspect it's competitive with any other method you might
>devise in the context of a flow for displaying something to the user.

Performance isn't my concern, it's just the complexity of all the implicit
behavior going on mapping the various pieces together to get a page
rendered that I can render myself with a handful of lines. Also the
concern noted below about supporting external templates.

>I'm curious about the interest in Velocity for rendering views. Is it
>primarily standardization on one templating system for numerous use
>cases (configuration templates was mentioned recently), or are
>there other reasons (i.e. performance)?

No, the reason is to avoid having the UI inside the war tree where it
can't be modified easily or, in the case of packed wars, dynamically. And
unpacked wars cause no end of confusion for people so I dislike them
intensely and don't want key functionality to depend on that.

If people want to use JSP, that's fine, but I don't want anything in this
release to require it. If you're planning to heavily customize the page
with Java logic accessing the IdP internals, JSP is certainly much better,
but if you're not doing that, Velocity's externalization advantage is huge.

But I can see that because of the need for both, using the views will
probably be necessary to align the models and allow for both options. But
my concern is whether all the view handling will allow for externally
located templates. I'm concerned they may do their own template lookup
instead of using Velocity ResourceLoaders, but I don't know that they do.

-- Scott