Bug 13122 – std.algorithm.cartesianProduct output type immutability

void main() { import std.algorithm: cartesianProduct; immutable a = "ABC"; foreach (t1; cartesianProduct(a, a)) pragma(msg, typeof(t1)); immutable b = "ABC"d; foreach (t2; cartesianProduct(b, b)) pragma(msg, typeof(t2)); } The latest dmd 2.066beta3 outputs: Tuple!(dchar, dchar) Tuple!(immutable(dchar), immutable(dchar)) Expected output: Tuple!(dchar, dchar) Tuple!(dchar, dchar)

Hmm. Isn't this expected behaviour? After all, typeof(a) == immutable(dchar)[], so shouldn't the cartesianProduct result contain the same element type, i.e., immutable(dchar)?

Sorry, I meant, typeof(b).

(In reply to hsteoh from comment #1) > Hmm. Isn't this expected behaviour? Probably yes, but I have opened a discussion thread: http://forum.dlang.org/thread/[email protected]

(In reply to bearophile_hugs from comment #3) > (In reply to hsteoh from comment #1) > > Hmm. Isn't this expected behaviour? > > Probably yes, but I have opened a discussion thread: > http://forum.dlang.org/thread/[email protected] Don't want to bump the thread. The reason filtering a string yields dchar instead of immutable(dchar) is because string.front returns a decoded dchar by value. On the other hand iterating a dstring returns the dchar by ref, so you see its immutability. Shall we close this bug?

(In reply to Peter Alexander from comment #4) > The reason filtering a string yields dchar instead of immutable(dchar) is > because string.front returns a decoded dchar by value. On the other hand > iterating a dstring returns the dchar by ref, so you see its immutability. > > Shall we close this bug? The current behaviour is is a little strange, but I guess you are right. I close this down then, as invalid.

Yet another unfortunate consequence of autodecoding. :-(