[erlang-questions] erlang *****

Benjamin Tolputt bjt@REDACTED
Wed Mar 19 01:27:06 CET 2008

I was going to stay out of this one (it being about Erlang internals 
rather than Erlang distribution methods - my current interest); but I 
have to pipe up on this one.

Any language that provides me with syntax/semantics (as opposed to 
library functions) that I cannot estimate a time for is most definitely 
a "Bad Thing". As Richard rightly points out, I can look at the 
Fibonacci example you provided and SEE the issue there.

However, as I understand it, the pattern-matching change being asked for 
(i.e. the one with backtracking) is a nasty idea as I cannot tell 
anything about it's performance without know which one of the many 
possible matching methods the current Erlang VM uses to establish it. 
Combined with fact that there is an alternate proposal that satisfies 
the requirements (as I see them) with known performance metrics and that 
pattern-matching is atomic (meaning long matches stall other processes) 
and I would rather this NOT be implemented.

Oh, and the prize for most eye-catching statement of the day goes to....

Richard A. O'Keefe wrote:
> but the student decided to get a sex change instead of a Masters.


More information about the erlang-questions mailing list