Bug 7978 – UTF exception (memory corruption?) using std.range.chain (behavior changes with -g)

Compiling the following code using DMD 2.059 without -g results in a UTF exception (memory corruption?) after outputting ["a", "b",. Compiling with -g produces no exception but the output is incorrect (["a", "b", "0", "1", "2"] should be ["a", "b", "a0", "a1", "a2"]). Now, if you replace the iota line with with the commented out sequence line it works the other way. Without -g gives no exception (but correct output this time), with -g results in a UTF exception. It's a bit unsettling that adding debug info can change the behavior of the program. import std.stdio, std.range, std.algorithm, std.string; void main() { writeln(gen(["a", "b"]).take(5)); } auto gen(string[] lst) { auto prefix = lst[0]; auto a = iota(10).map!(a=>format("%s%d", prefix, a))(); //auto a = sequence!"n+1"().map!(a=>format("%s%d", prefix, a))(); return chain(lst.save, a); } Similar code can also produce access violations as can be seen here: http://forum.dlang.org/post/[email protected] I'm not sure how to go about reducing this. Sometimes I can get it to work (using a global fixed it in one case though wasn't suitable for my needs). Moving it out of its own function into main() also fixes it sometimes but not always.

I see my error. Prefix is local so can't be captured in the closure. Passing lst by ref and making it not an rvalue resolves this.

According to TDPL (section 5.8.1) it should be valid to reference local variables from a closure. So this is a valid bug. I'll let you decide whether to reopen it; I already opened issue 8832 for probably the same bug. (Looks like in both cases it's the map delegate that's causing trouble.)

I've reopened so I'm reminded to check on this if issue 8832 gets fixed.

Tested with 2.060 on DPaste: 32 bits 64 bits. DMD PASS(1) FAIL GDC FAIL PASS(2) LDC FAIL FAIL (1) output: ["a", "b", "0", "1", "2"] (2) output: ["a", "b", "\x01\x000", "\x01\x001", "\x01\x002"]) http://dpaste.dzfl.pl/fork/2cd77316

(In reply to comment #4) > Tested with 2.060 on DPaste: > > 32 bits 64 bits. > DMD PASS(1) FAIL > GDC FAIL PASS(2) > LDC FAIL FAIL > > (1) output: ["a", "b", "0", "1", "2"] > (2) output: ["a", "b", "\x01\x000", "\x01\x001", "\x01\x002"]) > > http://dpaste.dzfl.pl/fork/2cd77316 I'd just like to add that the PASS cases are still incorrect (obviously since the output differs). They do build though which is what someone reading this should take "PASS" to mean. I just tried with DMD git (2.061 alpha-ish at the moment) and it throws this error message now: Error: function std.range.chain!(string[],MapResult!(__lambda2, Result)).chain.Result.save cannot get frame pointer to gen So at least the compiler seems to be somewhat aware of a frame pointer problem now although it should be able to accept it by creating a closure for this bug to be fixed.

This is fixed in 2.062 (tested on Windows and Linux).

Fixed but no associated pull -> worksforme. Don't blame the messenger, that's how Walter wants it.