Ytukyg

Here's some code:

matches =  //([a-z]+)/(d+)/state/([a-z]+)/.match(address) # line 1

puts matches[0]?                                               # line 2

puts matches.try &.[0]                                         # line 3

matches datatype is (Regex::MatchData | Nil) according to typeof. Yet, line 2, referring to matches[0] fails with a compilation error. And I don't understand line 3 at all!

Could someone clarify?

Line 2: As you say, the type of matches is (Regex::MatchData | Nil). In case it happens to be nil, it wouldn't have the #? method, which makes the type checker angry. You are supposed to check whether match succeeded first:

matches =  //([a-z]+)/(d+)/state/([a-z]+)/.match(address)

if matches

  puts matches[0]?

end

Inside if, the type of matches is just Regex::MatchData (as we eliminated the Nil possibility), and the type checker can rest peacefully.

If you are sure your string will match, you can pacify the type checker with not_nil!, but that opens up a possibility of a runtime error, if your confidence in your data's conformance was unfounded:

puts matches.not_nil![0]?

Line 3: #try will perform the block, except when the caller is nil, when it returns nil. No guards are needed because #try is explicitly defined on Nil (as well as on Object).

It uses the shortcut syntax for blocks, where &.[0] is kind of equivalent to { |x| x[0] }.