#'
would be a prefix that meant "special" or "keyword". So reserved words
or language constructs would be represented as identifiers that start
with '#', e.g. #class, #enum, #for, #if, #while, etc.Operators seemed special, so I also decided that operators would have the same prefix:
#+, #==, #>>, etc. I've changed my mind, though, for two reasons:
- I no longer think operators are so special. LES and some other languages have an infinite number of potential operators and they're treated identically to normal functions, apart from their syntax.
- No language I know of uses the specific prefix '
#' - Most importantly, the syntax for naming operators turned out to be cumbersome: because an operator name like
#+generally cannot be used directly as an identifier in most languages, some sort of escaping mechanism is needed. In LES and EC# you have to write, in general,@`special identifier`(not just`special identifier`because the latter is treated as an operator, not an identifier). Therefore the identifier#+is written@`#+`(@#+also works in LES), which is a bit cumbersome and (more importantly IMO) creates a teaching challenge: I have to explain both what#is for and also what@and the backticks are for, and the student must remember not to mix up@and#. Plus, when someone encounters LES or EC# code for the first time it'll be harder to guess what@`#+`means compared to@`+`, and Google won't help ("Your search - @`#+` - did not match any documents.")
operator+, Haskell (+), and Nemerle @+. In all of these cases the additional characters near + can be understood as an escaping mechanism, so that the actual identifier name is simply +.Therefore, I will soon dedicate a day or two to removing the hash characters from the beginning of all operators in both LES and EC#. Possibly, I'll remember to update the documentation too. We'll see.
No comments:
Post a Comment