On Thu, Mar 26, 2009 at 05:45:12PM -0500, Mark Copper wrote:
> On Sun, Mar 22, 2009 at 11:04:15PM -0500, Dave Rolsky wrote:
> > On Sun, 22 Mar 2009, Mark Copper wrote:
> >
> > > Begging your forebearance with this silly question. It seems to be
> > > related to hanging Apache processes.
> > >
> > > I am occasionally getting the warning:
> > > Use of uninitialized value in concatenation (.) or string at
> > > /usr/share/perl5/HTML/Mason/Request.pm line 1191.
> > >
> > > On my Debian install, line 1191 is the concatenation line in the print
> > > sub:
> > >
> > > sub print
> > > {
> > > my $self = shift;
> > >
> > > # $self->{top_stack} is always defined _except_ in the case of a
> > > # call to print inside a start-/end-request plugin.
> > > my $bufref =
> > > ( defined $self->{top_stack}
> > > ? $self->{top_stack}->[STACK_BUFFER]
> > > : \$self->{request_buffer}
> > > );
> > >
> > > # use 'if defined' for maximum efficiency; grep creates a list.
> > > for ( @_ ) {
> > > $$bufref .= $_ if defined; # this is line 1191
> > > }
> > >
> > > $self->flush_buffer if $self->{autoflush};
> > > }
> > >
> > > My question: what can possibly be in @_ that satisfies the conditional
> > > "if defined" and still sets off the "unitialized value" warning?
> >
> > If $$bufref itself is not defined, that could do it.
> >
> > I'm not sure how that would happen in just rare cases. I'd think if it
> > happened it would happen often. I'd also think it would happen for
> > everyone else.
> >
> >
> > -dave
>
> This had me a bit flummoxed. On one hand, $$bufref is frequently not
> defined, but the "uninitialized value" warning is quite rare. And on the
> other, I didn't know the LHS of a concatenation could trigger the
> warning:
>
> $ perl -we'my $a; my $b=\$a; my $c="Help!";$$b .= $c; print "$a\n"'
> Help!
> $
>
> and certainly not the RHS:
>
> $ perl -we'my $a="Help!"; my $b; $a .= $b if defined $b; print "$a\n"'
> Help!
> $
>
> I just never thought of checking if both were not defined:
>
> $ perl -we'my $a; my $b; $a .= $b if defined $b; print "$a\n"'
> Use of uninitialized value in concatenation (.) or string at -e line 1.
>
> $
>
> Oh my...
>
> Mark
There is actually a moral to this story for people at the base of the
learning curve like me.
The diagnostics pragma didn't provide new info but did provide a bit of
advice: the warning may actually be inherited from the caller.
Helpful info was obtained using Carp as warnings handler, cf. Practical
mod_perl (example 21-1 in my copy but available on-line).
With the guilty component isolated, judiciously placed warnings isolated
the problem.
Mark
------------------------------------------------------------------------------
_______________________________________________
Mason-users mailing list
Mason-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/mason-users
