Imported From: https://issues.scala-lang.org/browse/SI-9930?orig=1
Reporter: Moses Nakamura (mosesn)
Affected Versions: 2.11.8
See #8476

@SethTisue said:
duplicate of SI-8476?

Moses Nakamura (mosesn) said:
@seth: not sure. I wouldn't consider a feature flag that turns off this warning a fix, for example.

@som-snytt said:
IMHO, f"x$$y", which emits literal "x$y", is not weirder than any other syntax. It's nicer to have a specially-named interpolator that also refuses to do any interpolation, so you don't make a mistake the other way. I like the name lit"x$$y" for that. It would error on lit"x$$y$z" if z happened to be a value.

Another idea is an interpolator that does char replacement, a"x@y". Or a macro literally("x$y") that rewrites to f"x$$y".

Any sort of filter for reporting would require syntax. An annotation might work for syntax, but also require more work like the dotty reporting initiative.

Moses Nakamura (mosesn) said:
lit"x$$y$z" would work for me, as long as it didn't error if z happened to be a value.

@som-snytt said:
That's the same as s"x$$y$z".

Moses Nakamura (mosesn) said:
I'm not sure I understand. Doesn't s"x$$y$z" convert to "x$ywhateverisinz"? What I would want is for lit"x$$y$z" to convert to "x$$y$z".

@SethTisue said:
Moses: #8476 is about the same problem, so it seems better to me to have a single ticket for discussing possible solutions.

@som-snytt said:
Just to clarify, any"$z" will error out if there is no value z; even if any is a macro, it doesn't get a chance to ignore the error and reparse the string to have arbitrary meaning, IIUC.

So, I suggested lit"x$$y" to mean "x$y" and in particular for the mistaken lit"x$$y$z" to error, because lit"x$$y$$z" was intended. It's just a way to cope with expressing constant strings with dollars, while still warning about "x$y".

I noticed that "interpolation" has the anagram "Operation: Lint".

Moses Nakamura (mosesn) said:
@seth: that's fair.
@ap: ugh, OK.