Tera version 2

[schungx]

It is a significant issue that Tera has moved to version 2.0, and to my horror the original 1.0 docs website has been taken down as well. All that's left is 2.0.

I am not sure what plans there are for Loco wrt Tera. Will Loco 1.0 move to Tera 2.0?

In that case I'd have to plan ahead because I have a pretty massive Tera installation and 2.0 is COMPLETELY DIFFERENT. In fact, the author says it is a complete rewrite.

Would you consider adding a tera2 feature? Or would Loco stick with version 1.0 for the time being?

Beware that users will not be able to get docs on 1.0 easily...

[jondot]

Thanks @schungx — here's the plan: Loco will move to Tera 2.0 in an upcoming 1.x release. Since 2.0 is a rewrite with meaningfully different template syntax, we'll ship it as a clearly-documented change with a migration guide and a prominent CHANGELOG callout, rather than carrying both engines side by side behind a tera2 flag (two divergent syntaxes over the same view/mailer layer isn't a clean story).

So for planning: you can stay on Tera 1.x today, and there'll be a defined migration point in a 1.x minor — I'll make sure the upgrade guide spells out the syntax deltas that matter for a large install like yours, so you're not left reverse-engineering it from the (now sparse) 1.x docs. Tracking that work on this issue. Appreciate you raising both the version question and the docs situation.

[schungx]

Thanks for this great framework. Although can't say the same for Tera... IMHO it is a bit fickle and the scripting language is hard to use and full of traps.

Then this completely non-backwards-compatible version 2, a complete rewrite that broke all existing templates. And pulling the v1 docs instead of maintain LTS.

I have once thought about just building an alternative over Rhai itself, which shouldn't be very difficult since a template is nothing more than a script with arbitrary text mixed in.

Maybe I can make an alternative ViewRenderer based on that in rhai-loco...

Do you think it'd be worthwhile?

PS. Now I sorta understand why many languages have their own templating system, like Java with JSP, .NET with ASP, Python has a few. That's because it is almost trivial to build one by just hooking into the parsing process.

PPS. There seems to be already a bunch of templating engines based on Rhai. The latest one is https://github.com/intentee/poet

Maybe one of them can be repurposed.