The arguments for startsWith are back-to-front.

[robertmassaioli]

Just try and read these two statements:

> "hello" `startsWith` "hello world"
true

And:

> "hello world" `startsWith` "hello"
false

I would expect the opposite results. It seems to break the standard pattern in Haskell and Purescript whereby the infix notation should read legibly as a sentence.

[kritzcreek]

Another "standard pattern" in Haskell or PureScript is that the data argument is the last to make composition and chaining with partial application nice:

all (startsWith "hello") ["hello world", "hello universe"]

So while your point is definitely valid it's not as clear-cut as you make it sound. I prefer the trade-offs the current API makes.

[robertmassaioli]

The all notation could be written as:

all (`startsWith` "hello") ["hello world", "hello universe"]

Which would still be readable and looks the same in both applications. What would be your approach to removing the reading confusion in the infix invocation of this method?

I understand your position, it is valid, but I do think this style is cleaner.

[kritzcreek]

That's not valid PureScript though.

The PS equivalent:
all (_ `startsWith` "hello") ["hello world", "hello universe"]
doesn't look as nice imo.

In purescript-strings this is handled by wrapping the non-data argument in a Pattern constructor, thus removing the ambiguity between the two arguments.

stripSuffix :: Pattern -> String -> Maybe String

[hdgarrood]

Agree with @kritzcreek: in my mind, the "data argument comes last" pattern is much more useful and well-established than the "infix notation reads like a sentence" pattern.

[menelaos]

@robertmassaioli Thank you for your interest in this library!

You are right that unfortunately infix notation does not read like a sentence. However, I am inclined to leave the argument order for startsWith as it is due to the reasons mentioned by @kritzcreek and @hdgarrood. In the Haskell world I could point you to https://hackage.haskell.org/package/MissingH-1.4.0.1/docs/Data-String-Utils.html#v:startswith which makes the same trade-off.

I could offer two potential solutions:

• Expose the same function as isPrefixOf which would make it more suitable for infix use. The function name would also be familiar to Haskell users.

  > "hello" `isPrefixOf` "hello world"
  true
  
  > "hello world" `isPrefixOf` "hello"
  false

• Add a new module Data.String.Utils.Infix which (among others) would expose startsWith with the arguments flipped.

Please let me know what you think!

[bklaric]

I realize I'm late to the party, but I exclusively write "hello world" # startsWith "hello" for readable chaining.

[hadronized]

For what it’s worth, I had the issue with this function today as I was expecting the same order as in Haskell with isPrefixOf. Bumping a major version of the library for that seems overkill, but if you ever have to make any breaking change, I think it would make sense to swap the order.