Engine performance

[raymank26]

Given a com.hubspot.jinjava.Jinjava#renderForResult method it seems that every time a template is rendered the String template is parsed using JinjavaInterpreter#render->parse methods each time.

Could you clarify why the rendering routine doesn't use some sort of a TemplateId->Node cache to reuse the parsed AST in the rendering step? Are there any obstacles to implement that?

[jasmith-hs]

This is something that someone can make themselves using and caching JinjavaInterpreter#parse and calling JinjavaInterpreter#render. But it would be interesting to explore making that within com.hubspot.jinjava.Jinjava too

[proxo]

If reusing of Interpreter/Node is recommended for performance gain, than how to safely replace bindings before a next render ?
Sth like interpreter.getContext.putAll(newBindings) ?