digitalmars.D.bugs - [Issue 1084] New: lazy variadic parameters break in strange way
- d-bugmail puremagic.com (31/31) Mar 28 2007 http://d.puremagic.com/issues/show_bug.cgi?id=1084
- d-bugmail puremagic.com (31/31) Mar 28 2007 http://d.puremagic.com/issues/show_bug.cgi?id=1084
- d-bugmail puremagic.com (12/12) Apr 04 2007 http://d.puremagic.com/issues/show_bug.cgi?id=1084
- d-bugmail puremagic.com (16/16) Aug 04 2009 http://d.puremagic.com/issues/show_bug.cgi?id=1084
http://d.puremagic.com/issues/show_bug.cgi?id=1084
Summary: lazy variadic parameters break in strange way
Product: D
Version: unspecified
Platform: All
OS/Version: All
Status: NEW
Severity: normal
Priority: P3
Component: DMD
AssignedTo: bugzilla digitalmars.com
ReportedBy: default_357-line yahoo.de
Consider the following code.
import std.stdio;
char[] test() { writefln("Eval test"); return "bla"; }
void test2(lazy char[][] bla ...) { foreach (elem; bla) writefln(elem);}
void main() { test2(test, test); }
It compiles (verified gdc .21/dmd .175 by h3 [I couldn't find anybody with
1.010; volunteers appreciated]), but when run, outputs
Eval test
Eval test
bla
<garbage>
Some observations:
- returning "bla".dup doesn't help.
- The garbage seems to be an array with the same length as whatever was
returned from test. I did not check to see where the pointer goes. Maybe that
might offer some clue.
If this has already been fixed in 1.010, sorry for wasting your time.
greetings&stuffies --downs
--
Mar 28 2007
d-bugmail puremagic.com
http://d.puremagic.com/issues/show_bug.cgi?id=1084
------- Comment #1 from fvbommel wxs.nl 2007-03-28 14:57 -------
What seems to be happening here:
The parameter list of test2 is "lazy char[][] bla ...".
The type of the parameter is an array of char[]s. Because of the "..." this
array is implicitly constructed on the stack when the function is called.
However, because of the "lazy" this implicitly construction is put into a
delegate that is then pushed onto the stack as the actual parameter.
The problem with this is that the delegate constructs an array on the stack and
then returns _that_ array.
So the "lazy" delegate is returning a reference to data on the stack (in it's
own stack-frame). This is always a Bad Idea.
(I guess combining "lazy" and "..." is not a good idea)
What you may have meant was:
---
/// A call constructs an array on the stack and pass a reference.
/// Passed "parameters" are evaluated before calling.
void test2(char[][] bla ...) {
foreach (elem; bla) writefln(elem);
}
---
or:
---
/// A call constructs an array of delegates on the stack and pass a reference
to that.
/// Passed "parameters" aren't evaluated until used.
void test2(char[] delegate()[] bla ...) {
foreach (elem; bla) writefln(elem());
}
---
--
Mar 28 2007
http://d.puremagic.com/issues/show_bug.cgi?id=1084
thomas-dloop kuehne.cn changed:
What |Removed |Added
----------------------------------------------------------------------------
Version|unspecified |1.010
------- Comment #2 from thomas-dloop kuehne.cn 2007-04-04 15:36 -------
Added to DStress as
http://dstress.kuehne.cn/run/l/lazy_03_A.d
http://dstress.kuehne.cn/run/l/lazy_03_B.d
http://dstress.kuehne.cn/run/l/lazy_03_C.d
http://dstress.kuehne.cn/run/l/lazy_03_D.d
--
Apr 04 2007
http://d.puremagic.com/issues/show_bug.cgi?id=1084
Don <clugdbug yahoo.com.au> changed:
What |Removed |Added
----------------------------------------------------------------------------
Keywords|wrong-code |accepts-invalid
CC| |clugdbug yahoo.com.au
--- Comment #3 from Don <clugdbug yahoo.com.au> 2009-08-04 07:31:14 PDT ---
This isn't wrong-code, the bug is in the user's code. Changing to
accepts-invalid.
(Perhaps lazy []... should be illegal?)
On DMD2, it displays:
bug(12): Error: escaping reference to local __arrayArg502
So it's a D1-only issue.
--
Configure issuemail: http://d.puremagic.com/issues/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
Aug 04 2009