Templating engine

[hayssams]

We currently use jinjava as our template engine.
I am wondering if we should switch to pebble

[hayssams]

I feel that we should switch to pebble for the following reasons:

1. Pebble is highly compatible with jinja and offer a similar set of features
2. Switching would not break compatibility with existing projects
3. The pebble website provide a very good understanding of the templating language
4. has even more stars than the jinjava project

[tiboun]

I didn't know pebble, it seems to be a better choice than jinjava !

[zedach]

@hayssams I agree with 1. 2. For 4. even if it's a fact, the jinjava project seems to be active (last release in august) and maintained by hubspot. For 3. it seems that peeble is more adapted to a spring ecosystem, 𝜇s based on spring boot. I vote for jinjava because it seems to me that the major difference is in the extensibility part which can be very usefull. Jinjava has several extension points for custom functionality, including custom tags, filters, expression tests, and functions, which provide a high degree of flexibility for developers / starlake users. I haven't found the equivalent of the extensibility part in the pebble documentation so please don't hesitate to correct me if I'm wrong

[hayssams]

@zedach Yep you’re right that jinjava is still maintained and looks pretty complete. Pebble actually covers all the features you mentioned, see the links below:

• Pebble expression test: https://pebbletemplates.io/wiki/operator/is/
• Filtering: https://pebbletemplates.io/wiki/filter/abbreviate/
• New tags & functions: https://pebbletemplates.io/wiki/guide/extending-pebble/

Spring support is available in a side package and not included in the core pebble package we intend to use.

[zedach]

@hayssams thanks for the informations! Pebble therefore seems a very good choice. OK for me.

[hayssams]

While migrating to Pebble, I noticed a regression that we cannot accept. Replacing a variable by another variable at runtime, makes us loose the end of line and completely break the indentation of the yml file.

I suggest that for the time being we keep using HubSpot and send users to the Jinja documentation since jinjava is really close to it.