digitalmars.D - [GSoC] Strange struct destructor-postblit-writeln interaction bug.
- Cristi Cobzarenco <cristi.cobzarenco gmail.com> Jun 22 2011
- dsimcha <dsimcha yahoo.com> Jun 22 2011
--bcaec520ef999f971804a64e04e4
Content-Type: text/plain; charset=ISO-8859-1
While working on my project, I ran into a very hard to trace bug that I
finally managed to isolate and reproduce on a small scale.
The code:
import std.stdio;
import std.conv;
struct Test {
int x;
this( int m ) { x = m; }
this( this ) { writeln( "Postblit: ", x ); }
~this() { x = 42; }
string toString() { return to!string( x ); }
}
int main(string[] argv) {
auto a = Test(3);
writeln( a );
return 0;
}
Prints 42 on my computer, instead of 3. The writeln in the postblit
constructor shows that the object is copied 6 times, and only on the last
copy x becomes 42 instead of 3. Somehow the destructor gets called on an
object before it is copied. The copying that causes the problem is in
format.d 1599 as far as I can tell with my debugger.
I don't know if it's high priority for everyone, but it certainly is for me.
---
Cristi Cobzarenco
BSc in Artificial Intelligence and Computer Science
University of Edinburgh
Profile: http://www.google.com/profiles/cristi.cobzarenco
--bcaec520ef999f971804a64e04e4
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
While working on my project, I ran into a very hard to trace bug that I fin=
ally managed to isolate and reproduce on a small scale.<div><br></div><div>=
The code:</div><div><br></div><div><div><div>import std.stdio;</div><div>
import std.conv;</div><div><br></div><div>struct Test {</div><div><span cla=
ss=3D"Apple-tab-span" style=3D"white-space:pre"> </span>int x;</div><div><s=
pan class=3D"Apple-tab-span" style=3D"white-space:pre"> </span></div><div><=
span class=3D"Apple-tab-span" style=3D"white-space:pre"> </span>this( int m=
) =A0 =A0 { x =3D m; }</div>
<div><span class=3D"Apple-tab-span" style=3D"white-space:pre"> </span>this(=
this ) =A0 =A0 =A0{ writeln( "Postblit: ", x ); }</div><div><spa=
n class=3D"Apple-tab-span" style=3D"white-space:pre"> </span></div><div><sp=
an class=3D"Apple-tab-span" style=3D"white-space:pre"> </span>~this() =A0 =
=A0 =A0 =A0 =A0 { x =3D 42; }</div>
<div><span class=3D"Apple-tab-span" style=3D"white-space:pre"> </span></div=
<div><span class=3D"Apple-tab-span" style=3D"white-space:pre"> </span>stri=
iv>
int main(string[] argv) {</div>
<div><span class=3D"Apple-tab-span" style=3D"white-space:pre"> </span>auto =
a =3D Test(3);</div><div><span class=3D"Apple-tab-span" style=3D"white-spac=
e:pre"> </span>writeln( a );</div><div><span class=3D"Apple-tab-span" style=
=3D"white-space:pre"> </span>return 0;</div>
<div>}</div></div></div><div><br></div><div>Prints 42 on my computer, inste=
ad of 3. The writeln in the postblit constructor shows that the object is c=
opied 6 times, and only on the last copy x becomes 42 instead of 3. Somehow=
the destructor gets called on an object before it is copied. The copying t=
hat causes the problem is in format.d 1599 as far as I can tell with my deb=
ugger.</div>
<div><br></div><div>I don't know if it's high priority for everyone=
, but it certainly is for me.</div><div><br></div><div><br clear=3D"all"><d=
iv><br></div><div>---</div>Cristi Cobzarenco<div>BSc in Artificial Intellig=
ence and Computer Science</div>
<div>University of Edinburgh<br>Profile: <a href=3D"http://www.google.com/p=
rofiles/cristi.cobzarenco" target=3D"_blank">http://www.google.com/profiles=
/cristi.cobzarenco</a></div><br>
</div>
--bcaec520ef999f971804a64e04e4--
Jun 22 2011
== Quote from Cristi Cobzarenco (cristi.cobzarenco gmail.com)'s article--bcaec520ef999f971804a64e04e4 Content-Type: text/plain; charset=ISO-8859-1 While working on my project, I ran into a very hard to trace bug that I finally managed to isolate and reproduce on a small scale. The code: import std.stdio; import std.conv; struct Test { int x; this( int m ) { x = m; } this( this ) { writeln( "Postblit: ", x ); } ~this() { x = 42; } string toString() { return to!string( x ); } } int main(string[] argv) { auto a = Test(3); writeln( a ); return 0; } Prints 42 on my computer, instead of 3. The writeln in the postblit constructor shows that the object is copied 6 times, and only on the last copy x becomes 42 instead of 3. Somehow the destructor gets called on an object before it is copied. The copying that causes the problem is in format.d 1599 as far as I can tell with my debugger. I don't know if it's high priority for everyone, but it certainly is for me.
This is definitely a pretty severe bug and I can reproduce it. Please file a bug report in Bugzilla. (http://d.puremagic.com/issues/) It's ok to occasionally bring up very high priority bugs on this newsgroup, but they should **always** be filed in Bugzilla as well because otherwise they get forgotten.
Jun 22 2011
Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> On 6/22/11 10:13 AM, dsimcha wrote:== Quote from Cristi Cobzarenco (cristi.cobzarenco gmail.com)'s article--bcaec520ef999f971804a64e04e4 Content-Type: text/plain; charset=ISO-8859-1 While working on my project, I ran into a very hard to trace bug that I finally managed to isolate and reproduce on a small scale. The code: import std.stdio; import std.conv; struct Test { int x; this( int m ) { x = m; } this( this ) { writeln( "Postblit: ", x ); } ~this() { x = 42; } string toString() { return to!string( x ); } } int main(string[] argv) { auto a = Test(3); writeln( a ); return 0; } Prints 42 on my computer, instead of 3. The writeln in the postblit constructor shows that the object is copied 6 times, and only on the last copy x becomes 42 instead of 3. Somehow the destructor gets called on an object before it is copied. The copying that causes the problem is in format.d 1599 as far as I can tell with my debugger. I don't know if it's high priority for everyone, but it certainly is for me.
This is definitely a pretty severe bug and I can reproduce it. Please file a bug report in Bugzilla. (http://d.puremagic.com/issues/) It's ok to occasionally bring up very high priority bugs on this newsgroup, but they should **always** be filed in Bugzilla as well because otherwise they get forgotten.
Please put [GSoC] in the title. Andrei
Jun 22 2011