REPL feedback

[Robert Field]

On 11/27/14 02:23, Roy van Rijn wrote:
> After some more hours of playing with the actual REPL environment,
> I've come across some unlisted bugs/TODO items.
>
> It seems to be fairly easy to get stuck in an endless
> ReplParser.parseCompilationUnit loop, for example when you forget
> 'class':
>
> -> A {
>>> int i = 0;
>>> }
> *endless loop here, never reaches EOF in while loop*

Ah, bad handling when the initial syntax is bad.  Fixed and pushed.
Thanks!

>
> And testing the parsing, the following (valid Java) isn't correctly parsed:
>
> -> int[] i = {1,2,3};
> |  Error --
> |  illegal start of expression
> |  int[] i = {1,2,3};
> |            ^
> |  Error --
> |  not a statement
> |  int[] i = {1,2,3};
> |             ^^
> |  Error --
> |  ';' expected
> |  int[] i = {1,2,3};

Interesting.  Initial parse is correct, but initialization is separated 
and these initializers can't appear in assignments, Working on it.

> |
>
> Is this kind of testing appreciated and/or needed at this point in
> time, and are patches or tests welcome?

This kind of testing is very much appreciated!
We are working on the API design.  When we have a proposal we will be 
sending that out.
Once we have API, the code will be radically restructured, so patches 
may not be that useful, at least now.
Tests are very much appreciated as well.

Thanks much,
Robert