non-strict mode being very strict with matching unmatched parens

[fommil]

since a recent melpa upgrade, I've been getting errors in non-strict scala-mode. An example file looks like this (point is at ^^^, which is just an indicator) and then type " to close off "bootstrap

ensimeScalaCompilerJarModuleIDs := {
  val v = (scalaVersion in LocalProject("bootstrap^^^)).value
  Seq(
    scalaOrganization.value % "scala-compiler" % v,
    scalaOrganization.value % "scala-library" % v,
    scalaOrganization.value % "scala-reflect" % v,
    scalaOrganization.value % "scalap" % v
  ).map(_ % org.ensime.EnsimePlugin.EnsimeInternal.name intransitive ())
}

but I get this stacktrace instead 😢

Debugger entered--Lisp error: (wrong-type-argument char-or-string-p nil)
  sp--escape-region(("\"") 138 139)
  sp-escape-open-delimiter()
  sp--post-self-insert-hook-handler()
  self-insert-command(1)
  funcall-interactively(self-insert-command 1)
  call-interactively(self-insert-command nil nil)
  command-execute(self-insert-command)

Anything obvious broken here or should I go investigate when it bothers me enough?

[Fuco1]

I can't repro this. Would you mind testing this on the latest master?

The result I get in both strict and non-strict mode is an insertion of \"

[fommil]

I can confirm I still get it with the latest stable release, I'll try melpa unstable later today.

[fommil]

yes, confirming I still get it on melpa unstable.

Hmm, must be something in my config then.

[Fuco1]

What version of scala-mode do you have? Apparently mine is ancient (scala-mode-20141205.1251). I'll upgrade that and try again.

[fommil]

I always run the most recent version of scala-mode. My config is public https://github.com/fommil/dotfiles/blob/master/.emacs.d/init.el#L611-L746

[Fuco1]

After upgrading I get the behaviour described in OP.

The quote character before bootstrap has "punctuation" syntax, which is something quite odd and I think might be the issue. In the old version it had string syntax which is why it worked.

Oh how much I hate font-lock syntax hacks :)

[fommil]

thanks for the workaround! I think our scala-mode regexes really need an overhaul. They can also be quite slow sometimes. Somebody wrote some good ones for Sublime but I just haven't had the motivation to look into it.