[Bioperl-l] Bio::Tools::GuessSeqFormat unknown return value?

[Andreas Kahari]

On Wed, Dec 10, 2003 at 09:35:55AM +0000, Heikki Lehvaslaiho wrote:
> 
> Bio::Tools::GuessSeqFormat returns now string 'unknown'. My question is: how 
> strongly should we enforce the convention that unknown values return undef?
> 
> In practice, the choises are (1) undef, (2) empty string, (3) zero, (4) 
> predefined string value.  What are are the advantages and disadvantages of 
> this. Discuss.
> 
> I seem to remember that Hilmar felt strongly about this. ;-)

I chosed to return "unknown" when the module wasn't able to
come up with one single good guess because I thought it would
be better than to default to a valid format, such as "fasta"
(this was the way of bioperl-1.2.3, see the new() method of
e.g. Bio::SeqIO).

In retrospect, I think undef would probably be a better value.


Here's a patch for it:

Index: GuessSeqFormat.pm
===================================================================
RCS file: /home/repository/bioperl/bioperl-live/Bio/Tools/GuessSeqFormat.pm,v
retrieving revision 1.2
diff -u -r1.2 GuessSeqFormat.pm
--- GuessSeqFormat.pm   2003/12/04 13:32:35     1.2
+++ GuessSeqFormat.pm   2003/12/10 10:03:05
@@ -50,7 +50,7 @@
 The guess() method of a Bio::Tools::GuessSeqFormat object will
 examine the data, line by line, until it finds a line to which
 only one format can be assigned.  If no conclusive guess can be
-made, the "unknown" format is returned.
+made, undef is returned.
 
 If the Bio::Tools::GuessSeqFormat object is given a filehandle
 which is seekable, it will be restored to its original position
@@ -383,8 +383,7 @@
  Function   : Guesses the format of the data accociated with the
               object.
  Returns    : A format string such as "swiss" or "pir".  If a
-              format can not be found, the "unknown" format will
-              be returned.
+              format can not be found, undef is returned.
  Arguments  : None.
 
     If the object is associated with a filehandle and if that
@@ -502,7 +501,7 @@
         # Seek to the start position.
         $fh->setpos($start_pos);
     }
-    return ($done ? $fmt_string : "unknown");
+    return ($done ? $fmt_string : undef);
 }


Cheers,
Andreas

-- 
|(--)|      Andreas Kähäri                                |-][-|
|-)(-|      EMBL, European Bioinformatics Institute       |[--]|
|(--)|      Wellcome Trust Genome Campus, Hinxton         |-][-|
|-)(-|      Cambridge, CB10 1SD                           |[--]|
|(--)|      United Kingdom                                |-][-|