[Bioperl-l] For CVS developers-potentialpitfall with "return undef"

[David Messina]

To look for problems related to Heikki's "return undef" sweep, I ran  
'make test' on both today's version of bioperl-live and on an older  
version I had checked out on May 12. This was done on OS X 10.4.6 and  
perl 5.8.6.


Here are the results:

Failures in today's version of bioperl-live but NOT in 5/12 version
===================================================================
- psm.t -
The psm.t error appears to be new, so the changes made to Bio/Matrix/ 
PSM/SiteMatrix.pm, particularly the one in _uncompress_string, may  
need to be examined.

Here's the error message:
Illegal division by zero at t/psm.t line 147, <GEN1> line 36.
t/psm........................dubious
         Test returned status 255 (wstat 65280, 0xff00)
DIED. FAILED tests 29, 32-48
         Failed 18/48 tests, 62.50% okay


Failures in 5/12 version of bioperl-live but NOT in today's version
===================================================================
- OntologyStore.t -
Neither OntologyStore.t or Bio/Ontology/OntologyStore.pm have been  
touched between 5/12 and today.

The error looks like a transient network problem to me, but I'm not  
sure:
-------------------- WARNING ---------------------
MSG: [1/5] tried to fetch http://cvs.sourceforge.net/viewcvs.py/ 
*checkout*/song/ontology/so.definition?rev=HEAD, but server threw  
500.  retrying...
---------------------------------------------------
[REPEATED 5 times -Dave]

t/OntologyStore..............FAILED tests 3-6
         Failed 4/6 tests, 33.33% okay


- RepeatMasker.t -
Jason made commits to RepeatMasker.t and Bio/Tools/RepeatMasker.pm  
between 5/12 and today, so this appears to be not 'return undef'- 
related.

- SeqVersion.t -
The SeqVersion error was due to a failure to find and load  
Bio::DB::SeqVersion::gi, which Brian O. noticed and corrected between  
5/12 and today, so this is not 'return undef'-related.



All the other test failures appear in both versions of bioperl-live,  
so presumably they are not affected by the 'return undef' changes.

Comparing with Paul Boutros's earlier report, I saw errors in DBCUTG  
that he didn't. Those only pop up when I run the optional remote- 
server tests, however. Perhaps Paul didn't run those and that  
accounts for the discrepancy?

Also, he saw errors in Biblio.t, Repeatmasker.t, and  
StandAloneBlast.t that I did not.

Dave


Today's bioperl-live test results:

Failed Test        Stat Wstat Total Fail  Failed  List of Failed
------------------------------------------------------------------------ 
-------
t/Annotation.t                   89    2   2.25%  79 88
t/DBCUTG.t                       29    5  17.24%  26 30-32
t/LocusLink.t                    23    1   4.35%  23
t/PhysicalMap.t                  14    2  14.29%  11-12
t/TaxonTree.t                    17   30 176.47%  11 18-42
t/alignUtilities.t                9    1  11.11%  9
t/psm.t             255 65280    48   35  72.92%  29 32-48
t/tutorial.t                     21   15  71.43%  7-21
114 subtests skipped.
Failed 8/232 test scripts, 96.55% okay. 18/11057 subtests failed,  
99.84% okay.

Note that this is including tests requiring a remote server.

And here's the output from a May 12 checkout of bioperl-live:

Failed Test        Stat Wstat Total Fail  Failed  List of Failed
------------------------------------------------------------------------ 
-------
t/Annotation.t                   89    2   2.25%  79 88
t/DBCUTG.t                       29    5  17.24%  26 30-32
t/LocusLink.t                    23    1   4.35%  23
t/OntologyStore.t                 6    4  66.67%  3-6
t/PhysicalMap.t                  14    2  14.29%  11-12
t/RepeatMasker.t                  6    3  50.00%  1-2 6
t/SeqVersion.t      255 65280     6   10 166.67%  2-6
t/TaxonTree.t                    17   30 176.47%  11 18-42
t/alignUtilities.t                9    1  11.11%  9
t/tutorial.t                     21   15  71.43%  7-21
114 subtests skipped.
Failed 10/232 test scripts, 95.69% okay. 12/11049 subtests failed,  
99.89% okay.




On Jun 7, 2006, at 12:03 PM, Paul Boutros wrote:

> Hi,
>
> Just did a make test on bioperl-live on AIX 5.2.0.0 and perl 5.8.7  
> and I had a few
> failures:
>
> Failed Test         Stat Wstat Total Fail  List of Failed
> ---------------------------------------------------------------------- 
> ---------
> t/Annotation.t                    89    2  79 88
> t/Biblio.t                        24    1  2
> t/LocusLink.t                     23    1  23
> t/PhysicalMap.t                   14    2  11-12
> t/RepeatMasker.t                   6    3  1-2 6
> t/StandAloneBlast.t               18    4  19-22
> t/TaxonTree.t                     17   30  11 18-42
> t/alignUtilities.t                 9    1  9
> t/psm.t              255 65280    48   35  29 32-48
> t/tutorial.t                      21   15  7-21