On 9/18/07, Duncan Coutts
I want to start a discussion about a string searching api, especially with respect to ByteStrings.
My use case for searching bytestrings was to determine where a particular substring occurred in a search string. Thus, I needed the index. Taking the length of a string returned in a tuple would give me that index, but would feel dirty. I've seen DonS post enough to NOT take "length" that I wouldn't have thought to do that. I also wonder if the "tupling" style would make it easier to produce space leaks, especially if the "overlapping" behavior caused the tail of each string to be included. I don't quite understand sharing and the bytestring implementation to know how much of a concern that would be. I have to agree with Chris that I prefer findSubstrings to return a list of Int/Int64s, rather than the tuples. Justin