On 30.12.2013, at 10:44, Marcus Denker <marcus.den...@inria.fr> wrote:
>
> On 28 Dec 2013, at 14:25, Tobias Pape <das.li...@gmx.de> wrote:
>
>>
>> On 28.12.2013, at 10:07, Stéphane Ducasse <stephane.duca...@inria.fr> wrote:
>>
>>> I guess that
>>>
>>> #: should be #’:’
>>
>> That is one way.
>> In this particular case,
>> self property asMutator
>> was the better version, tho.
>>
>> Best
>> -Tobias
>> PS: Just curious, what was the reason to no longer allow #: ?
>> Or, where can I find the respective discussion, just want to lern.
>
> Pharo3 uses the new compiler infrastructure by default. This means that
> instead of the old (hand-written) parser, we now use RBParser (hand written,
> too).
>
> It seems that RBParser does not implement it. The question is who was wrong:
> the old
> parser or the new? I have no idea.
> But the good news is that we now just have two Parsers, not three… (syntax
> highlighting implements it’s
> own parser, too).
>
> What we need is a real grammar… that can solve these differences which come
> from the implementation.
Ansi has one, but I think it is unpractical, dividing “symbols” and “quoted
selectors” :(
>
> Another example is #9 —> both RB and the old compile it as an integer, there
> is special code in the Parser
> for that even. But do we want that? If we would have a definition of our
> grammar, we could look there and
> say “yes, it’s part of the definition”.
>
> Even better would be an executable grammar… we should explore of we can use
> PetitParser in the future,
> but there are some open questions (error handling, speed…).
>
> The most beautiful would be to have just *one* parser that has a high level
> model of it’s grammar that is
> reusable everywhere.
Hehe. I think, both PP and Ometa would fit that criterion but I agree
that this is hard-think for the core-Parser of a Smalltalk :)
Best
-Tobias
signature.asc
Description: Message signed with OpenPGP using GPGMail
