[Bioperl-l] How to handle bugs in bioperl 1.4 on CPAN?

[Chris Fields]

On Jun 25, 2006, at 2:02 PM, Phillip SanMiguel wrote:

> Chris Fields wrote:
>> On Jun 24, 2006, at 2:37 PM, Phillip SanMiguel wrote:
>>
>> [...]
>>> How long before the next "stable" release? Maybe a year? Should  
>>> not a
>>> BioPerl 1.4.1 be released so CPAN would get bug fixes like this  
>>> one? Or
>>> would that be very difficult?
>>
>> No, it's not that easy.  BioPerl isn't like most CPAN modules with  
>> one or two developers.  See the wiki page for details on planning  
>> releases to see why:
>>
>> http://www.bioperl.org/wiki/Making_a_BioPerl_release
>>
>> It takes a lot of effort and coordination, much more so than the  
>> average CPAN module.  I believe some of the core developers are  
>> meeting this weekend; maybe something will come of that and we'll  
>> get an idea of a next release.
>>
>> Chris
> Hi Chris,
>   Thanks for the information--the key part being that a bug fix  
> from a couple of years ago has not propagated into the current  
> stable release. Below I'll try to convince you that this is a  
> serious problem. (Not because it is your fault, of course. I'm just  
> trying to deliver my take on the situation to the bioperl- 
> programmer-warriors who happen to be listening...)
>   It isn't a problem for me to edit the offending statement in the  
> QualI.pm module on systems I generally use. Or even install a  
> developer's release of bioperl. My problem is one of advocacy.  
> Maybe I have a warped view of the world, but it seems that except  
> for those directly involved in the bioperl or GMOD projects,  
> everyone looks to CPAN when they install bioperl.

Again, it's not as easy as you make it seem.  The idea is to upgrade  
the CPAN version to stable releases (even numbered) and that odd- 
numbered releases would be developer versions.  Yes, it has been a  
while since the last stable version; it could be a while until the  
next as there have been suggestions of an interim 1.5.x release or so  
before that occurs (though he did say 1.6 could be soon after BOSC  
which is in August).  Hilmar has explained that there are some  
stumbling blocks to get around before the next major release (if  
those 'stumbling blocks' are what I think they are, I agree).  It's  
very likely implementation of changes that he mentions may require  
refactoring code, changing API, etc.  Not easy in a project like  
this, a large core of contributors and with the developers scattered  
all over the world, all with different priorities (we all have $jobs  
after all).

That's why we have a Release Pumpkin, akin to the Pumpkings that have  
ushered forth regular perl releases.  It requires a large,  
coordinated effort with one person acting as overseer, pushing  
everybody to meet deadlines.  Not easy and not, by a long shot, your  
typical CPAN module.

>    I write scripts that I sometimes want to send to biologists even  
> less programming-capable than I am. I can just barely envision  
> those biologists pestering their sysadmin to do a CPAN install of  
> bioperl modules so that my script will work. But installing a non- 
> CPAN set of modules probably isn't going to happen.
>    So, this being the case, how can I, with a clear conscious,  
> advocate bioperl to the junior bioinformaticians with whom I happen  
> to interact?

Give those biologists some credit. Quite frankly, I would expect any  
bioinformaticist or computational biologist, junior or otherwise, to  
know or at least learn how to install from CPAN or from CVS,  
otherwise they need to change their job title.  And, as a  
microbiologist myself (i.e. one of those biologists you mention) and  
as one who regularly interacts with biologists with little to no  
computer science experience, I believe I can speak from experience.   
I find the install documents that come with BioPerl and available on  
the wiki pretty much cover everything, from how to install to the  
dependencies required to problems one may encounter.   The web site  
has a tone of documentation, including the FAQ (*cough* which covers  
this ground *cough*).

If they are running perl scripts and using a system that requires  
sysadmin privileges they probably know what thy are doing anyway.  If  
not they probably have students/employees that do know what's going  
on (and who may be the ones actually running the scripts).  You can't  
please everybody, so I think you can proceed with a clear conscious  
knowing you did the best that you can to help!

>    My take, for what it is worth, is that 1.5 has become an  
> unratified stable release. How hard would it be to take 1.5.1--as  
> is--and deposit that in CPAN? What would be the downside?

Ah I see Hilmar has responded.  I think he adequately answers this.   
API is everything; changing API suddenly is bad bad bad.

> Phillip SanMiguel

Christopher Fields
Postdoctoral Researcher
Lab of Dr. Robert Switzer
Dept of Biochemistry
University of Illinois Urbana-Champaign