[Bioperl-l] Bug fixes; SeqIO issues

[Hilmar Lapp]

On Monday, September 2, 2002, at 01:48  AM, Ewan Birney wrote:

>
> Many thanks to the bug fixes by Steve/Jason and others
>
>
> I have commited some more bug fixes from the bug tracking list. 
> Slow and
> steady is my aim here.
>
>
> SeqIO has probably currently a nasty hidden bug due to fuzzy loading. I
> think this is due to heikki's move to atomic locations which he 
> wanted to
> handle indels more cleanly. However, something is broken down there -
> silently!

> I have an exception being thrown but have switched this off until I can
> chat with Heikki about the route out of here.
>

The problem here is no problem. You tried to check arguments for 
$start and $end for being numeric in a constructor. This is a bad 
idea because it breaks extension by inheritance -- derived classed 
may (and in this case in fact do) allow values the base class didn't 
dream of. What happens is that start() and end() in the fuzzy class 
parse the values.

If you really want to enforce in the base class that start() and 
end() return numeric values, then set first and check the return 
values (they will indeed be numeric). However, if someone sets not 
through constructor but calling the setters you can't control this 
anymore.

Am I missing something? Anyway, I removed the check ...

>
>
> Yes - cue discussion of how messed up our-location-or-seqfeature 
> heirarchy
> is and I don't know the right route out of here. Hmph.
>

I'm not that unhappy with it I have to admit ... although I'm sure 
there's lots of room for improvement. What's bothering you?

	-hilmar

--
-------------------------------------------------------------
Hilmar Lapp                            email: lapp at gnf.org
GNF, San Diego, Ca. 92121              phone: +1-858-812-1757
-------------------------------------------------------------