[Bioperl-l] Bio::Root::IO _readline/_pushback behavior

[Chris Fields]

Yeah, it's a queue; the 'buffering' is a simple internal array using push/shift.  I say we merge the change in from the branch and fix any modules accordingly.

chris

On Aug 4, 2011, at 4:31 PM, Hilmar Lapp wrote:

> I agree. In fact I'm surprised that $io->_pushback() does not act like unshift() - that's I thought how it is used.
> 
> 	-hilmar
> 
> On Aug 3, 2011, at 10:16 AM, Chris Fields wrote:
> 
>> On Aug 3, 2011, at 3:55 AM, Kai Blin wrote:
>> 
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>> 
>>> Hi folks,
>>> 
>>> as I mentioned on https://redmine.open-bio.org/issues/3264 there is
>>> something odd going on with Bio::Root::IO's _readline/_pushback
>>> functions. This seems to be intentional, at least there is a test case
>>> asserting the behaviour I'm seeing. It his however very confusing to the
>>> unexpecting programmer using the code.
>>> 
>>> One assumption I'd immediately make would be that if I have code that
>>> does a $foo = $io->_readline; $io->_pushback($foo); $bar =
>>> $io->_readline;, $foo will be the same string as $bar, regardless what
>>> other pieces of the code did. Currently, this is not the case, because
>>> the readbuffer that _pushback pushes back into has new strings appended
>>> to the end but readline removes them from the front.
>> 
>> I think this test is performed in the regressions already, but if not then it is more than welcome.
>> 
>>> This easily violates the "principle of least surprise", so I think we
>>> should change the readbuffer to a stack. As far as I can tell, changing
>>> the _pushback function to "unshift" instead of "push" to the readbuffer
>>> breaks only the Root/RootIO.t test designed to test the old behaviour. I
>>> don't see any other tests failing on my system that don't fail without
>>> this patch.
>>> 
>>> Any comments from the core devs?
>> 
>> I don't have a problem with that beyond the change to the RootIO.t tests (it implies a specific behavior that some developers expect, so is a very subtle API change).  However, this is how one would expect it, to be more like an 'unread' stack instead of a queue.  In fact, there is a module I used for Biome's pushback/readline called IO::Unread that implements an IO layer for mimicing this behavior, might be worth looking into.
>> 
>>> Cheers,
>>> Kai
>> 
>> chris
>> 
>> 
>> Christopher Fields
>> Senior Research Scientist
>> National Center for Supercomputing Applications
>> Institute for Genomic Biology
>> University of Illinois Urbana-Champaign
>> 1206 W. Gregory Dr. , MC-195
>> Urbana, IL 61801
>> 
>> 
>> _______________________________________________
>> Bioperl-l mailing list
>> Bioperl-l at lists.open-bio.org
>> http://lists.open-bio.org/mailman/listinfo/bioperl-l
> 
> -- 
> ===========================================================
> : Hilmar Lapp -:- Durham, NC -:- hlapp at drycafe dot net :
> ===========================================================
> 
> 
> 
> 
> _______________________________________________
> Bioperl-l mailing list
> Bioperl-l at lists.open-bio.org
> http://lists.open-bio.org/mailman/listinfo/bioperl-l