On Mar 28, 2011, at 5:16 PM, Ketil Malde wrote: > Nicholas Ingolia <n...@ingolia.org> writes: > >> I'm pretty strongly in favor of 0-based indexing, > [...] >> If you're committed to 1-based indices, > > Three votes for zero so far, and I'm not very committed, so I guess zero > it is. But > >> we could exploit the newtype wrapper around Offset to provide an >> interface with 1-based offsets. > > this line of thought is interesting, I think. Perhaps we can supply an > Offset1 newtype as well for use with 1-based data formats? Probably > only internally, so that e.g. a SeqLoc will always use zero-based > offsets.
I was thinking that there could be an alternate Bio.SeqLoc.Position1Based module that re-exported Offset as an opaque data type and provided mkOffset1 :: Int64 -> Offset unOffset1 :: Offset -> Int64 The internal representation would be unchanged but the client would see the conversion of offsets to numbers using 1-based indexing (and your call on how it deals with the -1/+1 jump). Handling the Num instance would be tricky, some of the operations would be wrong (e.g., literal 0 of type (Num a) => a would not be shifted properly) and others would be confusing (e.g., you don't add two 1-based offsets) but I don't believe we can control the re-export of the Offset Num instance. Best, --Nick _______________________________________________ Biohaskell mailing list Biohaskell@biohaskell.org http://malde.org/cgi-bin/mailman/listinfo/biohaskell
