[p2pu-dev] Tech call notes: 2013-01-17
chris at p2pu.org
Thu Jan 24 05:24:09 UTC 2013
@Dirk - so there's a few reasons why I don't like the django templating
layer. First off, I am spoiled with erb which lets you use the full ruby
language in the templates. I think that using a full language rather than
tags or whatever is usually preferable - it's more powerful and is one less
del to learn. The downside is that they usually a little more verbose. On
the JS side, I know that ejs and eco work similarly, giving you full JS in
your templates, so my preference is for that as opposed to mustache or
handlebars. The filters is what really bug me with the django templates. I
don't really like the syntax for them and it feels to me like a hodgepodge
of theoretically useful functions in a global context... it's approaching
PHP style which is really bad IMHO. I don't know all of the filters, so I'm
never really sure if I should do string manipulations in the controller
function or with a filter. It's also not straightforward how you define and
use your own filters - to me at least, maybe this is not true. All of this
is a non-issue if you are using a full programming language in your
templates where you can create and use non-global (even OO - yay!)
functions the same way as the rest of the language.
I'm not attached to ejs in any way really, but I do think it makes sense to
do the view layer of this app in JS. As we mentioned - the big advantage of
using a JS view layer is that they all speak JSON, so we wouldn't need to
change anything from our REST API. Freebie!
@José - I shared with you the original grant proposal for the MOOC Maker.
Check it out, but read it skeptically. It's the super long term version -
right now we want to make each MOOC it's own instance so we can learn from
the experience of running different MOOCs. Really right now, all we want is
to recreate what we currently have in ruby in python and add an email
scheduler. The app described in the proposal is pretty big and would
definitely be a django app.
As for JS MVC frameworks - I have experience with backbone and JSMVC.
Definitely prefer backbone. JSMVC is huge, complex, and obtrusive with way
too many moving parts to understand it all. Definitely do not recommend it.
Backbone is the polar opposite - the bare minimum you need to get a MVC up
and running. And it's dependency underscore.js is really nice too -
basically a very well thought out core lib extension. I was up and running
with backbone in 1 day. The code is really easy - but the hardest part is
the way of thinking with it - where a 'view' is always associated with a
DOM element and a Collection is basically an array stack of those views,
with a bunch of callbacks for adding / removing them. Backbone also has a
biggest user base / community of the JS MVCs by a large margin so that's a
really good sign IMHO.
I haven't played around with AngularJS at all. How do you like it? I think
it's also pretty lightweight? Also, do you have an opinion of JS templating
projects that you like / dislike?
Also, I did some quick googling of flask. Check this out:
Apparently our line of reasoning is not original. Let me know what you
On Wed, Jan 23, 2013 at 8:36 AM, Dirk Uys <dirk at p2pu.org> wrote:
> Hi Jose
> On Tue, Jan 22, 2013 at 7:28 PM, Jos Flores <josmasflores at gmail.com>wrote:
>> If all your business logic is encapsulated in classes, your domain
>> objects will be usable by both Django and flask, and being plain
>> objects, can be very testable. But anyway, this is pretty much the
>> same you are saying! :)
> I tend to agree with you on this. I prefer writing domain logic behind a
> stateless API that answers questions. Anything behind your API is free to
> use any data store. Resources should be referenced by URI and you have the
> possibility of weak references. Exposing an APIs over HTTP follows easily.
> p2pu-dev mailing list
> p2pu-dev at lists.p2pu.org
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the p2pu-dev