Re: Problems with seq 2.2.4 on Windows

[Andrew Johnson <anj@aps.anl.gov>]

On 01/13/2017 05:02 AM, freddie.akeroyd@stfc.ac.uk wrote:
> snl.c is autogenerated using the lemon parser and lemon was updated
> between 2.2.3 and 2.2.4
>  
> Looking at the code failing it is:
>  
>     }else{
>       assert( yyact == YY_ERROR_ACTION );
>       yyminorunion.yy0 = yyminor;
> #ifdef YYERRORSYMBOL
>       int yymx;
> #endif
>  
> I think Visual Studio 2010 doesn’t like the declaration of  yymx 
> occurring after an assignment, this wasn’t allowed prior to the C99
> standard I believe.

IIRC the vxWorks 5.5.x compiler will also object to the above source
code; K&R and ANSI C89-rules required all variable declarations to
appear before any other statement within a block, which the above code
violates when YYERRORSYMBOL is defined. GCC has allowed it for a long
time though, so some programmers might never have come across the rule.

Note that the rule applied within any {code block;}, it didn't require
all variable declarations to appear at the beginning of a subroutine
definition. A lot of older EPICS code was written that way, but I don't
know if that was due to misunderstanding the rule or just preferring
that particular coding style. I usually move variable declarations into
the inner-most scope possible when I work on such code nowadays.

- Andrew

-- 
Arguing for surveillance because you have nothing to hide is no
different than making the claim, "I don't care about freedom of
speech because I have nothing to say." -- Edward Snowdon