fcall legal in scanner?

[Carlos Antunes]

Thanks, Adrian! Will this fix make it into 5.15?

On 10/29/06, Adrian Thurston <thurs... at cs.queensu.ca> wrote:
> Hi Carlos,
>
> I found the bug. The problem was with fret in scanner actions. The
> adjustment to p that is needed in some cases was not happening. The
> attached patch will fix the problem.
>
> Cheers,
>   Adrian
>
> Carlos Antunes wrote:
> > On 10/27/06, Adrian Thurston <thurs... at cs.queensu.ca> wrote:
> >> It sounds like a bug. fcall is supposed to work in actions associated with
> >> entire patterns (the lex style actions) but they are illegal in the the
> >> transition/state based embeddings and Ragel is supposed to emit an error.
> >> Could you post or send me some code which exhibits the problem?
> >>
> >
> > Adrian,
> >
> > Please, see test2.rl file attached. It's a messy file because it's
> > where I do, testing.
> >
> > Basically, I'm using a scanner (main) to match 'Organization',
> > 'Authorization' and a general purpose header called extension_header.
> > If there is a header like 'an-extension-header', the scanner should
> > match on the extension_header rule.
> >
> > Here's a string to match:
> >
> >             "Authorization: Digest username=\"Alice\",realm=\"atlanta.com\"\n"
> >             "    , response=\"7587245234b3434cc3412213e5f113a5432\" , \r\n"
> >             "    nonce=\"84a4cc6f3082121f32b42a2187831a9e\" \r\n"
> >             "Organization: Boxes by\r\n Bob  \r\n      \n"
> >             "An-Extension-Header: 10 \n"
> >             "\r\n";
> >
> > The scanner performs well with the previous string. Let's try a new string:
> >
> >             "An-Extension-Header: 10 \n"
> >             "Authorization: Digest username=\"Alice\",realm=\"atlanta.com\"\n"
> >             "    , response=\"7587245234b3434cc3412213e5f113a5432\" , \r\n"
> >             "    nonce=\"84a4cc6f3082121f32b42a2187831a9e\" \r\n"
> >             "Organization: Boxes by\r\n Bob  \r\n      \n"
> >             "\r\n";
> >
> > In this case, the scanner, once it matches the extension header, will
> > continue to match 'Authorization' and 'Organization' as extension
> > headers as well, which didn't happen before.
> >
> > Note that, if instead of using fcall from the main scanner, I use
> > fgoto (and them fgoto back to main instead of fret), both input
> > strings produce the same, and correct, behavior.
> >
> > Thanks!
> >
> > Carlos
> >
>
>
> >
>
> Index: ragel/xmlcodegen.cpp
> ===================================================================
> --- ragel/xmlcodegen.cpp        (revision 3695)
> +++ ragel/xmlcodegen.cpp        (working copy)
> @@ -246,6 +246,9 @@
>         case InlineItem::Break:
>                 out << "<break></break>";
>                 break;
> +       case InlineItem::Ret:
> +               out << "<ret></ret>";
> +               break;
>         default: break;
>         }
>
> @@ -360,12 +363,9 @@
>                 case InlineItem::Goto: case InlineItem::GotoExpr:
>                 case InlineItem::Call: case InlineItem::CallExpr:
>                 case InlineItem::Next: case InlineItem::NextExpr:
> -               case InlineItem::Break:
> +               case InlineItem::Break: case InlineItem::Ret:
>                         writeWithContext( item, context );
>                         break;
> -               case InlineItem::Ret:
> -                       out << "<ret></ret>";
> -                       break;
>                 case InlineItem::PChar:
>                         out << "<pchar></pchar>";
>                         break;
>
>
>


-- 
"We hold [...] that all men are created equal; that they are
endowed [...] with certain inalienable rights; that among
these are life, liberty, and the pursuit of happiness"
        -- Thomas Jefferson