[Bioperl-l] bioperl-run parameter question

[Chris Fields]

On Dec 11, 2006, at 3:19 PM, Sendu Bala wrote:
...

>>
>> I respectfully disagree that this is a 'useful' distinction.  My main
>> point is consistency.
> [snip]
>
> We're on the same page in terms of what we think would be a Good  
> Thing,
> and allowing both ways (dashed and dashless) sounds reasonable. I was
> just suggesting why bioperl-run might be the way it was. Further to
> that, there is the practical aspect that it is a lot simpler to figure
> out which are the program options so they can be farmed out to the
> AUTOLOAD methods - again something that isn't done in core.

Maybe b/c AUTOLOAD is frowned upon for a number of reasons, mainly  
code maintenance.  I'm somewhat neutral on the idea of using AUTOLOAD  
as a short-term solution, though using heredoc and an eval{} block  
works well for me (and shows up when using $self->can('method') or  
when checking for methods via Class::Inspector).

> If you come up with some generic way of dealing with options and  
> farming
> to AUTOLOAD, perhaps there's scope for applying it to all the run
> wrappers (ideally via one of their base classes), so they all  
> instantly
> gain dashed-mode capability.

I think that's the crux of the problem; they do not all have the same  
base class (except Bio::Root::Root).  Most use WrapperBase.  I  
thought at one point a Run-specific root module would be a good idea,  
but WrapperBase already works well.

I'll go ahead with my modules and think about it some more.  You  
could ask the powers-that-be (jason, hilmar, etc) what they think as  
well.

chris