[Bioperl-l] Added test for multiple fasta parsing

Hilmar Lapp hlapp@gmx.net
Fri, 10 Nov 2000 09:24:39 -0800

Elia Stupka wrote:
> Hello folks,
> I am sorry we were panic-busy over here for a while.
> I have now double-checked, the fasta parser could not parse multiple
> entries on a perfectly normal fasta file, I have used the multifa.seq file
> in the test suite to check.
> I have now commented out that buggy check from the main trunk as well. I
> have also committed a multiple fasta parsing test, to prevent further bugs
> like this.

I am sorry for insisting on this, but I don't think we have found the
problem yet. I downloaded the test script you put into the repository,
without updating the rest (my SeqIO directory was last updated Oct 11th).
This test script prints

nereis-4: bioperl-live 8:56 124> perl t/multiple_fasta.t
ok 1
ok 2
ok 3
nereis-4: bioperl-live 8:56 125>

for both the 06-branch and the main trunk. So if I take the output of that
script as the indicator for the presence or absence of a bug in
multiple-entry fasta reading, as you suggested, then there was no bug.

This is at least true for Linux 2.2.10/Perl 5.003, and Irix 6.5/Perl 5.6
(which we use frequently and happily at work for reading fasta files with
100 times more entries than the test file).

So, there are two reasons I am insisting. The first is that I don't like if
people claim code as buggy and 'fix' it without proving that it really was
incorrect (fixing something that was correct in the worst case introduces a
new bug, and in the best case doesn't change anything), and the second is
that if there is a problem with the code (and there apparently was one for
Elia) and you 'fix' it at the wrong place, the actual problem may be
silenced (for a while?) but it is still there.

So, as I consider the fasta-parser as being central, could we please
carefully re-check what the problem was, which file revealed it, and under
which circumstances (OS, perl version) it occurred. Or simply put, could I
please have a normal and complete bug report that can be used for debugging
(as opposed to trial and error). I will be happy to even download a 10 or
100MB-sized fasta file (but don't attach more than 2-3 MB to an email,


Hilmar Lapp                                email: hlapp@gmx.net
GNF, San Diego, Ca. 92122                  phone: +1 858 812 1757