[alerts] investigate usage of handlebars instead of mustache

[pmuellr]

For alerting, we currently use mustache as our templating library to render strings with variables from the alert execution. We've wondered about switching to handlebars instead, and this issue is to discuss/research that.

Handlebars is basically mustache++. One significant issue we'd have to deal with is whether all our existing customer's templates would still work.

Beyond that, I did just a bit of peeking at handlebars for issue #79371 (comment) , and noted both the complexity issues (on the customer's end) and security concerns. I remember coming away from the investigation feeling that handlebars is very much aimed at providing developers a richer experience than mustache, which it certainly does. But we're not targetting developers - ideally it would be nice to have something in-between mustache and handlebars - a little more function, but not so much that the result is a new programming language.

What led us looking at handlebars again is being able to provide ease-of-use functions for use in a template. And the fact that Dashboard URL templating uses handlebars.

[elasticmachine]

Pinging @elastic/kibana-alerting-services (Team:Alerting Services)

[mikecote]

Here is the link to a prior conversation about using mustache instead of handlebars: #37043 (comment).

[pmuellr]

Some work was done in this PR to investigate: #146487

Handlebars didn't make it that much easier to add extension functions, and has two big downsides:

• users would have to opt-in to use it instead of Mustache
• the templating can actually get more complex with more error possibilities when using the "piping" capabilities

So, for now, this doesn't look like it's going to add useful value to customers.