On Fri, 27 Oct 2000, Jarkko Hietaniemi wrote:
> > [.ext.socket]socket.obj, []perlshr_bld.opt/Option, perlshr_attr.opt/Option
> > %LINK-W-NUDFSYMS, 1 undefined symbol:
> > %LINK-I-UDFSYM, S_SYYLEX
> > %LINK-W-USEUNDEFSYMV, undefined symbol S_SYYLEX referenced
> > in symbol vector option
> > %MMS-F-ABORT, For target PERLSHR.EXE, CLI returned abort status:
> > %X10648268.
> >
> > Was something unusual done with any /s_syylex/i symbols recently?
>
> Well, yes, S_yylex was *added*. It didn't exist before. Ignasi Roca
> suggested the concept and I tried to implement it: the symbol was introduced
> to split S_yylex() into a static (S_syylex) and "dynamic" part (S_yylex)
> parts to allow re-entrant parsing. Maybe I forgot something VMS-specific
> that has to be done when symbols are added?
Thanks I see it in toke.c. It looks like S_syylex made it into
the transfer vector, and that is where the problem currently lies:
$ search PERLSHR_BLD.OPT;1 syylex
SYMBOL_VECTOR=(S_syylex=PROCEDURE)
This will require a bit of messing with.
Peter Prymmer
