[m-rev.] for review: extend mutual tail rec optimization to functions

[Julien Fischer]

Hi Zoltan,

On Tue, 12 Sep 2017, Zoltan Somogyi wrote:

> Another attached file contains the output of a --grade java bootcheck
> that fails for a reason that seems to me to be completely unrelated
> to this diff. Julien, if you could take a look at it and tell me what
> you think is wrong, I would appreciate it. (The bootcheck was done
> on the machine you gave me access to,)

Bootchecking the Java (and indeed C#) grade in the past has required a
certain amount of manual intervention.  It looks like the issue here is
that the classpath is not correctly set for the stage 2 compiler.
(For Mercury Java grade executables the classpath is set by the wrapper
script we generate, I suspect that's not correct for the stage 2 compiler.)

I'll look into this one.

Incidentally, the machine I gave you access to now has a C# compiler
installed on it.

Julien.