[NewCompiler] Parser error alternateKeyword symbole?

[Marcus Denker]

On 18.09.2006, at 19:46, Mathieu Suen wrote:

> Hi,
>
> I want to generate a alternateKeyword symbole but I can find a way  
> to do it.
> In fact I dfon't know what is an alternateKeyword symbole.
>
> I pick it up from the grammar:
> ...
> |	"#" <alternateKeyword>	{#litSymbol:}
> ...
>
> And I want to build a test like:
>
> rbNode := (self defaultParser parseMethod: 'test #:indexOf:') body  
> statements first.
> 	self assert: (rbNode firstToken value = '#').
> 	self assert: (rbNode lastToken value = ':indexOf:')
>
> but (self defaultParser parseMethod: 'test #:indexOf:') raise an error
>
> Have an idea?
>

The old (handwritten) parser indeed accepts stuff like

1)
  #:hallo:hallo:

and

2)
  #hallo:hallo

Both of which are not parsable with the Smacc based parser.

I guess that the alternatKeyWord token was supposed to fix problem
1), but it does not work.

So this is a bug that needs to be fixed... or we need to decide to
only support well-formed symbols (#sdfsdf:sdfsdf:), as for the
others people can write #':hello:hello' with no problem.
I tend to the second solution, as we re-compile the image and
use lots of code with the new compiler, and this problem never
came up.

Does anyone have an example of those strange symbols in praktice?

      Marcus