[Biocorba-l] Part V - The final result (fwd)

[Ewan Birney]

Alan -

nice to see this post. 


One thing that came back from bioperl (from my perspective) is that the
SeqFeature interface with

   $feature->start 
   $feature->end

should still be (a) maintained and (b) not throw exceptions.


I think translating this to IDL means that if we go for something like


   $feature->location_root()

we still have to have 

   $feature->start
   $feature->end

The server *has to make a decision* about how to represent the location
tree below him. (not allowed to punt this one back to the client). If we
punt everything to the client, the interface becomes less useful because
the client has to wrap everything in complex descending/messing around
code etc etc...


I know. This means the server has to decide what to do with nasty
fuzzy-locations-across-multiple-sequences with numerous joins. But someone
has to make the call about how this "looks" on a linear sequence, and the
server has more knowledge than the client.


The interface should work well with the 99% use case, which is either


   (1..100)

OR

   (1..100,200..300,400..500)


making it not work well with the 1% fuzzy-nightmare-join is ok in my book.


Does this sync with your thoughts? Would you prefer to see something
different?




-----------------------------------------------------------------
Ewan Birney. Mobile: +44 (0)7970 151230, Work: +44 1223 494420
<birney@ebi.ac.uk>. 
-----------------------------------------------------------------