On Tue, 3 Oct 2006, Bonzini Paolo wrote:
> Well, you could say that you need only two instead of three: in my case it
> is "outside the parser class" (e.g. for imports) and "inside" (for
> methods used in the actions). But they don't really map to header/code,
> so I'm using %{...%} respectively before/after %%.
I agree that header vs. code is not the best terminology for this Java
issue.
> > 3. Change %start-header{...} to %type-code{...}. For C/C++ users, this is
> > the right place to put any code that is required by YYSTYPE or YYLTYPE.
> > You probably don't need it in Java, but you could use it if you'd like to
> > organize your output.
>
> I don't like the %type-code name (constructive criticism in a couple
> of paragraphs!).
How about %dependencies{...}? In Java, it could be used for your imports.
%code{...} could be used for your parser methods. %header{...} would have
no meaning and could be disallowed for Java, right?
> > 4. Remove %before-header entirely. We're not sure it's actually
> > necessary.
>
> I would put #include directives for the parser
> there in C (e.g. <stdlib.h>), since %before-header IIRC is just good
> old %{...%}. But in the end putting them in %after-header wouldn't be
> a problem, right?
That's my thinking.
> A renaming proposal working for me would be %after-header -> %code, and
> %before-header -> %prolog, and likewise %header-code and %header-prolog.
I think Akim would like to get rid of %before-header for C/C++, and I can
live without it as well.
To reduce confusion, I'd like not to use the word prologue since it
already refers to %{...%} and all its baggage.
> This, plus letting these appear in the grammar section, would be just
> perfect.
Letting them appear in the grammar section should already be possible in
CVS Bison. Whether they appear there or in the definitions section does
not affect what they do.
