« first day (369 days earlier)      last day (208 days later) » 

13:37
Do we ignore error output?
I noticed that SWI-Prolog just eats syntax errors. So if we allow error input it would be super easy to add.
 
2 hours later…
15:53
While it doesn't give a non-profit exit code, its debug output throws a message it describes as an error. The same occurred with INTERCAL I believe, and we didn't allow the error.
 
3 hours later…
18:30
I think such errors should be allowed. The rule about stderr output states that only fatal errors are not allowed. The important thing about error condition is whether it causes the program to terminate, not how it is called.
Strictly speaking, yash stderr output consists of errors: "cannot execute command" is clearly an error, yash (and other shells) just do not call it so.
It is a too weak distinction - how interpreter names an error condition, I think we should not rely on that.
For example, ><> interpreter prints "Something smells fishy" on all invalid commands - it also does not have t
 
3 hours later…
21:05
@stasoid I agree that the distinction is too weak, and was not a fan of the prior interpretation. STDERR it is.

« first day (369 days earlier)      last day (208 days later) »