port to Jinja2

[chadwhitacre]

Advantages over Tornado 1.x:

1. easier to debug (probably? provide better tracebacks for templating errors #1381)
2. supports i18n (i18n #1030)
3. provides super (IRC)

Granted, Tornado 1.x is ancient. However, 3.x may address (1) but afaict they don't address (2) nor (3).

[chadwhitacre]

Error-reporting is currently... uh, interesting.

Looks like they haven't even addressed (1) yet.

[zwn]

👍

[clone1018]

It seems like Jinja2's goal is to keep python out of templates. Which is does by rendering the templates in a new context, so we don't have any python functions out of the box, we have to "approve" specific ones we want.

I've made a semi complete comparison of a couple of templating engines I found for python.

[clone1018]

So far Mako looks pretty good but the syntax is different, Django also looks good.

[clone1018]

It doesn't look like Django's Templating Engine is modular though, so we wouldn't be able to easily pull it out for use with Aspen. Mako is it's own thing that can easily be included in our project.

[chadwhitacre]

I'm pretty sure we don't want embedded Python syntax.

What's wrong with whitelisting template-global functions?

[clone1018]

Global functions includes native Python functions. That's not a limitation of tornado.

And our template loops and calls look like python, but the Jinja the syntax is different.

On my phone so I will clarify later.

[clone1018]

Cool, we talked about it in IRC and decided on Jinja2

[zwn]

Anything missing here? I don't think so.