[katello-devel] Thinking about NuTupane Architecture

Tom McKay thomasmckay at redhat.com
Mon Jan 7 14:38:07 UTC 2013



----- Original Message -----
> From: "Jason Rist" <jrist at redhat.com>
> To: "Lukas Zapletal" <lzap at redhat.com>
> Cc: katello-devel at redhat.com
> Sent: Monday, January 7, 2013 9:24:31 AM
> Subject: Re: [katello-devel] Thinking about NuTupane Architecture
> 
> On Mon 07 Jan 2013 07:19:22 AM MST, Lukas Zapletal wrote:
> > Hmmmm I don't think this library is mature, I'd rather stay away
> > from
> > it. And how does this help us to avoid coding our own UI
> > components?
> > >From what I read, Backbone.js is complementary to jQuery and other
> > things.
> >
> > LZ
> >
> > On Mon, Jan 07, 2013 at 07:02:07AM -0700, Jason Rist wrote:
> >> On 01/07/2013 01:50 AM, Lukas Zapletal wrote:
> >>> On Fri, Jan 04, 2013 at 04:15:03PM +0000, Dmitri Dolguikh wrote:
> >>>> As long as the views do not deviate from the standard ones and
> >>>> do
> >>>> not require custom ui elements, the mixed (html+js) approach to
> >>>> developing views works quite well. However, it introduces some
> >>>> issues (in addition to the ones mentioned by Jason):
> >>>>  - oftentimes controllers require additional methods to
> >>>> retrieve/update a particular part of the view
> >>>>  - creating custom UI elements is time consuming
> >>>>  - testing of interactions between individual ajax
> >>>>  elements/calls is hard
> >>>>  - somewhat related to the one directly above - custom page/tab
> >>>> updates are hard (i.e. only specific UI elements need to be
> >>>> updated,
> >>>> the whole tab needs to be re-rendered, etc)
> >>>>  - page assembly is mostly defined by templates, which makes
> >>>> customization/testing hard
> >>>>
> >>>> What if we used pure js controllers and views (think backbone,
> >>>> or
> >>>> other "real" MVC approach) working against API controllers? This
> >>>> approach would potentially address quite a few of the
> >>>> issues/concerns above, and improve testability too.
> >>>>
> >>>> I suspect that we'd be able to support both approaches (the one
> >>>> currently used and pure js one) simultaneously, which eliminates
> >>>> the
> >>>> need for "one big rewrite".
> >>>>
> >>>> While I don't think we should plunge into this head-first, I do
> >>>> think the idea warrants a trial/spike. Thoughts/opnions?
> >>>
> >>> I can agree only if we would use some existing JS framework for
> >>> that. We
> >>> do not want to build UI components from scratch again. Do you
> >>> have
> >>> something specific on your mind?
> >>>
> >>> In general, I agree :-) +1
> >>>
> >>> LZ
> >>>
> >> I think Dmitri was suggesting that we use something like Backbone,
> >> which
> >> already exists.  Is that what you mean?
> >>
> >> -Jason
> >>
> >> --
> >> Jason E. Rist
> >> Senior Software Engineer
> >> Systems Management and Cloud Enablement
> >> Red Hat, Inc.
> >> +1.919.754.4048
> >> Freenode: jrist
> >> github/identi.ca: knowncitizen
> >>
> >> _______________________________________________
> >> katello-devel mailing list
> >> katello-devel at redhat.com
> >> https://www.redhat.com/mailman/listinfo/katello-devel
> >
> 
> I humbly disagree. Backbone has been around for several years and has
> had many iterations.  It's at 0.9.9 which is almost as close as you
> get
> to a one point oh.  We already use Underscore as a dependency so
> adding
> this would be trivial.  Aeolus Conductor already uses Backbone to
> provide a large number of UI syncing mechanisms so we have some areas
> of expertise to lean on as well.
> 
> -J
> 

I agree, backbone is solid from what I've read.

> --
> Jason E. Rist
> Senior Software Engineer
> Systems Management and Cloud Enablement
> Red Hat, Inc.
> +1.919.754.4048
> Freenode: jrist
> github/identi.ca: knowncitizen
> 
> _______________________________________________
> katello-devel mailing list
> katello-devel at redhat.com
> https://www.redhat.com/mailman/listinfo/katello-devel
> 




More information about the katello-devel mailing list