Binding a variable as the tail or not - reasons for this language design choice?

Questions / Help
,
1

SOLVED (how do I mark this as solved?): Turns out I had left behind junk in my repl/shell and R was already bound. Oopsie. Thanks for the help

Hi, I’ve just started learning erlang and using it for work after 15 years in other languages.
Like all new platforms and tools, some things are new and surprising coming from a different enviornment.

I’d like to learn the reasons for some language design choices, the first one I found really surprising being this one here:

5> case [a] of
   [] -> wtf;
   [A|R] -> a_r;
   [A] -> a
   end.
a

6> case [a] of
   [] -> wtf;
   [A|_] -> a_r;
   [A] -> a
   end.
a_r

Here we can see that the second pattern becomes semantically different if we bind a variable as the tail or not (R vs _)

I’m having a lot of fun so far with the langauge, but some things feel like they could be useful to have in a list of surprising behaviors coming from other backgrounds (Example: One book that was really useful learning the langauge “Go” was “100 go mistakes and how to avoid them”, which lists surprises in that langauge)

What would you recommend for learning the background to the language choice above?

2

Here we can see that the second pattern becomes semantically different if we bind a variable as the tail or not (R vs _)

Which Erlang interpreter are you using? The behavior you are reporting is not observable with OTP 27.2.

There is no semantic difference between _ and A (in your example), they are just identifiers. _ stands for a new unused identifier name each time it occurs, and the compiler never warns that it is unused.

3

When I run your first example, I get a_r as the result. This suggests to me that (to get a) you must already have R bound, meaning that the case clause doesn’t match.

Remember: in Erlang, names are bound to values once; they cannot be re-bound.

1 Like
4

Thanks for the replies. Indeed, it turned out that I had R already bound :smiley: in my shell. So it was all expected behavior.

Sorry for the mess, I was unable to revoke the original post before it went live, as it ended up with status pending (my first post).

Issue is solved!

3 Likes