[antlr-interest] Removing output=AST lets ANTLR fail to generate

[Eric]

On Fri, Nov 11, 2011 at 1:55 PM, Sam Barnett-Cormack <
s.barnett-cormack at lancaster.ac.uk> wrote:

> So, in other words, the feature request is that AST-related syntax be
> ignored if output!=AST?
>
> Yeah, I could go for that, though I don't know if there would be
> complications deeper in implementation. Any tree-related action code,
> for instance, would be broken horribly. However, it would be a useful
> thing for very clean, pure lexing and parsing. The only thing I wonder
> is, for that not-generating-an-AST-any-more parser, wouldn't it need
> some different actions from the original AST-generating version in order
> to be any use whatsoever?
>
> Sam
>
> On 11/11/2011 17:51, Christian wrote:
> > Waoh, your interpretation of my words really shocked me ;)
> >
> > I just asked if ANTLR could generate the lexer and parser without
> > AST-code if the option output=AST is removed. Apparently, it is not
> > possible so far. And please believe me, I was at no moment angry. But
> > back to the topic:
> >
> > Feature request for ANTLR v4:
> > *Context:*
> > If you add AST shaping code to a given grammar and later change the
> > grammar itself, too, you cannot directly generate a pure lexer and
> > parser without AST shaping code anymore. You have to remove the AST
> > shaping code from the grammar to do it.
> > *Solution:*
> > If a combined grammar contains AST shaping code, it would be more
> > flexible if ANTLR is still able to generate non-AST target language code
> > when, e.g., the option output=AST is not defined.
> >
> > I hope, my phrasing is better now ;)
> >
> > Best regards,
> > Christian
> >
> > Am 11.11.2011 16:33, schrieb Jim Idle:
> >> I think it might be the phrasing of your question. It may be a language
> >> barrier issue, but it sounds like "You stupid ANTLR asshats, why can't
> you
> >> do it this way, which is the way I expect it to work and so it must be
> the
> >> correct way; you are all stupid!"
> >>
> >> But...
> >>
> >> Clearly, as you get an error telling you that you have AST shaping code
> in
> >> there, then you cannot remove that option without removing the AST
> shaping
> >> directives. Whether or not that can be improved is not very relevant as
> >> 3.x is now in  maintenance mode while ANTLR 4 is being thrashed in to
> >> shape. So, perhaps you can rephrase your statement in to a feature
> request
> >> for ANTLR v4. It is a reasonable request if a little care is taken over
> >> the phrasing.
> >>
> >>
> >> Jim
> >>
> >>> -----Original Message-----
> >>> From: antlr-interest-bounces at antlr.org [mailto:antlr-interest-
> >>> bounces at antlr.org] On Behalf Of Christian
> >>> Sent: Friday, November 11, 2011 12:15 AM
> >>> To: Eric; antlr-interest at antlr.org
> >>> Subject: Re: [antlr-interest] Removing output=AST lets ANTLR fail to
> >>> generate
> >>>
> >>>> As I understand it, you are creating a grammar and then remove
> >>>> output=AST
> >>> I created a combined grammar that contains operators and rewrite rules.
> >>> When I then remove the option "output=AST", the grammar is not
> >>> generated due to syntax errors. These syntax errors result from the
> >>> operators and rewrite rules. You can reproduce it with any
> >>> corresponding grammar you like.
> >>>
> >>> Christian
> >>>
> >>> Am 10.11.2011 18:33, schrieb Eric:
> >>>> I don't believe that anyone here is avoiding your question. I my self
> >>>> am unable to directly answer your question because it does not make
> >>>> sense to me. As I understand it, you are creating a grammar and then
> >>>> remove output=AST. Since you are expecting a lexer and parser as
> >>>> output, I must take it that you  are using a combined grammar. Based
> >>>> on the limited info given I see no reason why you are not getting an
> >>>> lexer and parser. If you were to post your grammar and the steps you
> >>>> take, someone here might spot something needing advise.
> >>>>
> >>>> Eric
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>      >  Hi community,
> >>>>      >
> >>>>      >  it seems that ANTLR cannot generate the lexer and parser if
> >>>>      output=AST
> >>>>      >  is removed, but the operators and rewrite rules are still left
> >>>>      in the
> >>>>      >  grammar. Am I right? If yes, that's really regrettable...and
> >>>>      unflexible.
> >>>>      >  You could just ignore rewrite rules and operators, if output
> !=
> >>> AST,
> >>>>      >  couldn't you?
> >>>>      >
> >>>>      >  Regards,
> >>>>      >  Christian
> >>>>
> >>>>      List: http://www.antlr.org/mailman/listinfo/antlr-interest
> >>>>      Unsubscribe:
> >>>>
> >>>> http://www.antlr.org/mailman/options/antlr-interest/your-email-
> >>> address
> >>>>
> >>>
> >>> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> >>> Unsubscribe: http://www.antlr.org/mailman/options/antlr-interest/your-
> >>> email-address
> >> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> >> Unsubscribe:
> http://www.antlr.org/mailman/options/antlr-interest/your-email-address
> >>
> >
> >
> > List: http://www.antlr.org/mailman/listinfo/antlr-interest
> > Unsubscribe:
> http://www.antlr.org/mailman/options/antlr-interest/your-email-address
>
> List: http://www.antlr.org/mailman/listinfo/antlr-interest
> Unsubscribe:
> http://www.antlr.org/mailman/options/antlr-interest/your-email-address
>

After thinking about it, how I would like to see it implemented would be in
ANTLRWorks.

If you create a combined grammar that has lexer, parser, and AST rules, you
could select an option to refactor the grammar into three separate grammar
files.


Eric