Re: Debugging "$4"

[Maury Markowitz]

> On Oct 8, 2020, at 2:17 AM, Akim Demaille <akim@lrde.epita.fr> wrote:
> Chris might be very right here.  Maury, have you looked at your debug traces? 
>  
> https://www.gnu.org/software/bison/manual/html_node/Tracing.html

I turned on YYDEBUG (my duh) and it appears the parse is working as expected? 
Recall the line is:

200 IF 1=1 THEN PRINT"3":PRINT"4"

Relevant bits... Its parsed the "IF 1=1 THEN" part and notices the PRINT:

    Shifting token THEN ()
    Entering state 122
    Reading a token: Next token is token PRINT ()

Then after many rules fire on the the PRINT it gets:

    Next token is token ':' ()

Many more roles fire as a result, unwinding the stack and producing:

Reducing stack by rule 26 (line 364):
   $1 = token PRINT ()
   $2 = nterm printlist ()
-> $$ = nterm statement ()

So the first PRINT is now turned into a statement, as expected. Then it 
continues:

    Shifting token ':' ()
    Entering state 92
    Reading a token: Next token is token PRINT ()

A similar set of rules to the first print fires and ends here:

    Reducing stack by rule 26 (line 364):
       $1 = token PRINT ()
       $2 = nterm printlist ()
    -> $$ = nterm statement ()

The second print is also a statement now. The line is now complete and it reads 
the \n, so it unwinds the rest:

    Reducing stack by rule 8 (line 179):
       $1 = nterm statement ()
       $2 = token ':' ()
       $3 = nterm statements ()
    -> $$ = nterm statements ()

To me it *appears* that it has successfully fired the "statements" rule? 
Continues...

    Reducing stack by rule 21 (line 315):
       $1 = token IF ()
       $2 = nterm expression ()
       $3 = token THEN ()
       $4 = nterm statements ()
    -> $$ = nterm statement ()

That all looks correct to me, but I am noob.

I made a fatal mistake basing this on g_list, which uses a (void *) to the 
"data" (my statement_t) and thus debugging the C side is a PITA.