I changed all the generated code to use template strings because I think it makes it a little bit easier to see what's going on and make changes. The generated code is the exact same for all the tests, except for javascript where I added a missing semicolon and missing whitespace between an identifier and an &&, not that it really matters. One thing to note is that I dedent the multiline template string at runtime just to I can keep it two tabs over without messing up the generated code, which I think is better than the alternative. I also extracted out the common text in all the generated comments.
I know it's really just a matter of style. It seems that you avoid es6 stuff, which is a fine reason to reject this PR. I wasn't sure if the exclusion of "newer" JS features is still a deliberate choice today.
I changed all the generated code to use template strings because I think it makes it a little bit easier to see what's going on and make changes. The generated code is the exact same for all the tests, except for javascript where I added a missing semicolon and missing whitespace between an identifier and an &&, not that it really matters. One thing to note is that I dedent the multiline template string at runtime just to I can keep it two tabs over without messing up the generated code, which I think is better than the alternative. I also extracted out the common text in all the generated comments.
I know it's really just a matter of style. It seems that you avoid es6 stuff, which is a fine reason to reject this PR. I wasn't sure if the exclusion of "newer" JS features is still a deliberate choice today.