RE: odd recdescent 1.80 behaviour

Ok, this seems to have been caused by some interaction via sub AUTOLOAD 
in a module of mine which was subclassing P::RD. Hmm. It's really tricky 
tracing warnings sometimes!

Thanks very much for the help,
R.


On Wed, 29 Jan 2003, Orton, Yves wrote:

> > Does anyone have any idea why this works perfectly:
> > while(<FIN>) {
> >   print $parser->record($_);
> > }
> > 
> > but the following fails to parse exactly the same input:
> > 
> > while(my $rec = <FIN>) {
> >   print $parser->record($rec);
> > }
> 
> I dont think its relevent in this situation but the above two whiles arent
> equivelent you know.
> 
> while(defined(my $rec = <FIN>)) {
> 
> is equivelent to
> 
> while(<FIN>) {
> 
> > The latter throws the following warnings:
> > Use of uninitialized value in index at 
> > /usr/local/lib/perl5/site_perl/5.8.0/Parse/RecDescent.pm line 
> > 2979, <GEN0> line 67.
> > Use of uninitialized value in index at 
> > /usr/local/lib/perl5/site_perl/5.8.0/Parse/RecDescent.pm line 
> > 2979, <GEN0> line 67.
> > Use of uninitialized value in length at 
> > /usr/local/lib/perl5/site_perl/5.8.0/Parse/RecDescent.pm line 
> > 2734, <GEN0> line 67.
> > Use of uninitialized value in substitution (s///) at (eval 
> > 80) line 2207, <GEN0> line 67.
> > Use of uninitialized value in substitution (s///) at (eval 
> > 80) line 2207, <GEN0> line 67.
> > 
> > This would perhaps suggest that the input is different but 
> > it's definitely not. 
> 
> Er, since you didn't give us any sample data its a bit hard to tell if this
> is true. Sorry to doubt you but....
> 
> For instance it might be interesting to see the output of Devel::Peek on the
> two values.
> 
> Ive played with some of my P::RD's under these two constructs and I see no
> difference at least in 5.6.1. (Dont have 5.8 to check against right now)
> 
> > The same warnings are thrown if I read the same input 
> > data from a 
> > socket-server source.
> 
> Based on my experience with 5.6.1 and 5.8 I would think that this has to do
> with 5.8 new IO layers. Try changing it to use the perlio layers directly. I
> believe that you can do this easiest by setting the enviornment variable
> 
> PERLIO=perlio 
> 
> BTW, if this doesnt help then can you try to replicate this under 5.6.1 and
> if there is no error then report it along with as minimal a piece of code
> that you can that replicates the problem to 
> 
> [email protected]
> 
> It might be worth trawling the p5p archives and the bug reporting system to
> see if this is in fact a 5.8 error that has already been resolved/worked
> around etc.
> 
> A last thought is that it could have to do with what OS you are on. I know
> that red hat has problems sometimes due to UTF8 setting (this is wild
> guess). seeing the output of perl -V would be useful.
> 
> Anyway, I hope some of this turns out to be useful, let me know how it
> turned out,
> Cheers,
> yves
> 
> 
> 

-- 
+---------------------------------------------------------------+
 Roger Michael Pettett           Email: [email protected]
 Project Leader (Web Systems),   Web: http://www.sanger.ac.uk/
 The Sanger Institute                 http://www.yourgenome.org/
 Wellcome Trust Genome Campus         http://www.ensembl.org/
 Hinxton                       
 Cambridge CB10 1SA
+---------------------------------------------------------------+