Re: [BUG] line numbering off-by-one with Apache::DB and RegistryCooker

[prev] [thread] [next] [Date index for 2005/03/17]

Dominique Quatravaux wrote:
>>In which case you will end up with use warnings on the same line as the 
>>shebang line, no? which will look strange in the debugger
> 
> 
> Actually I find this pretty idiomatic, look:
> 
>    use warnings; #!/usr/bin/perl -w
> 
> No? :-) 

It may confuse some users who may claim that mod_perl has hijacked and 
rewritten their scripts :)

> Besides this line is seldom debugged anyway (at least in my
> experience) because the programmer typically sets a breakpoint with
> "$DB::single=1;" directly in his/her CGI.
> 
> The other problem is that I cannot Do The Right Thing while keeping
> the current API intact. I propose the attached patch as an alternative,
> but it rips the ModPerl::RegistryCooker->rewrite_shebang() method
> apart (hopefully nobody depends on it yet, as it is undocumented).

As 2.0 is not released yet we can still do minor changes. I doubt anybody 
has an overriden version of this particular API.

>>I wonder if it'd be cleaner to use the '#line number filename'
>>directive. 
> 
> 
> Actually this is what is currently done, but *before*
> ->rewrite_shebang() fiddles with the first line, which results in an
> off-by-one.

I like your solution, Dominique, since it's even faster than the current 
one as it avoids the:

   ${ $self->{CODE} } =~ s/^/$prepend/ if $prepend;

which forces a memory re-allocation to prepend a string (which can be 
pretty big).

Unless someone disagrees we will go with this change.

> Index: ModPerl-Registry/lib/ModPerl/RegistryCooker.pm
> ===================================================================
> --- ModPerl-Registry/lib/ModPerl/RegistryCooker.pm	(revision 157426)
> +++ ModPerl-Registry/lib/ModPerl/RegistryCooker.pm	(working copy)
> @@ -371,7 +371,7 @@
>      return $rc unless $rc == Apache::OK;
>  
>      # convert the shebang line opts into perl code
> -    $self->rewrite_shebang;
> +    my $shebang = $self->shebang_to_perl;
>  
>      # mod_cgi compat, should compile the code while in its dir, so
>      # relative require/open will work.
> @@ -397,6 +397,7 @@
>                      "sub handler {",
>                      "local \$0 = '$script_name';",
>                      $nph,
> +                    $shebang,
>                      $line,
>                      ${ $self->{CODE} },
>                      "\n}"; # last line comment without newline?
> @@ -553,13 +554,13 @@
>  }
>  
>  #########################################################################
> -# func: rewrite_shebang
> -# dflt: rewrite_shebang
> +# func: shebang_to_perl
> +# dflt: shebang_to_perl
>  # desc: parse the shebang line and convert command line switches
>  #       (defined in %switches) into a perl code.
>  # args: $self - registry blessed object
> -# rtrn: nothing
> -# efct: the CODE field gets adjusted
> +# rtrn: a Perl snippet to be put at the beginning of the CODE field
> +#       by caller
>  #########################################################################
>  
>  my %switches = (
> @@ -572,7 +573,7 @@
>     'w' => sub { "use warnings;\n" },
>  );
>  
> -sub rewrite_shebang {
> +sub shebang_to_perl {
>      my $self = shift;
>      my($line) = ${ $self->{CODE} } =~ /^(.*)$/m;
>      my @cmdline = split /\s+/, $line;
> @@ -588,7 +589,8 @@
>              $prepend .= $switches{$_}->();
>          }
>      }
> -    ${ $self->{CODE} } =~ s/^/$prepend/ if $prepend;
> +
> +    return $prepend;
>  }
>  
>  #########################################################################
> 
> 


        -- 
        __________________________________________________________________
Stas Bekman            JAm_pH ------> Just Another mod_perl Hacker
http://stason.org/     mod_perl Guide ---> http://perl.apache.org
mailto:stas@xxxxxx.xxx http://use.perl.org http://apacheweek.com
http://modperlbook.org http://apache.org   http://ticketmaster.com