zwang <nehzgnaw gmail.com> writes:

<code>
import std.stdio;
const char[] culprit;//#2
void f(...){	
     char[] s;
     void putc(dchar c){
         std.utf.encode(s, c);
     }
     std.format.doFormat(&putc, _arguments, _argptr);	
     //foreach(char ch; s){} //#9
     writef(s);
}
void main(){
     f("mystery");
}

</code>

The program above does not display "mystery" as expected.
The string can be displayed when line #2 is removed
or when line #9 is uncommented, both seemingly irrelevant.

I've smelt something fishy in dmd's code generator.

=?ISO-8859-1?Q?Thomas_K=FChne?= <thomas-dloop kuehne.THISISSPAM.cn> writes:

zwang wrote:

 <code>
 import std.stdio;
 const char[] culprit;//#2
 void f(...){
     char[] s;
     void putc(dchar c){
         std.utf.encode(s, c);
     }
     std.format.doFormat(&putc, _arguments, _argptr);
     //foreach(char ch; s){} //#9
     writef(s);
 }
 void main(){
     f("mystery");
 }

 </code>

 The program above does not display "mystery" as expected.
 The string can be displayed when line #2 is removed
 or when line #9 is uncommented, both seemingly irrelevant.

 I've smelt something fishy in dmd's code generator.

Can you reproduce this bug without using Phobo's functions?

Thomas

zwang <nehzgnaw gmail.com> writes:

Thomas Kühne wrote:
 zwang wrote:
 
 <code>
 import std.stdio;
 const char[] culprit;//#2
 void f(...){
     char[] s;
     void putc(dchar c){
         std.utf.encode(s, c);
     }
     std.format.doFormat(&putc, _arguments, _argptr);
     //foreach(char ch; s){} //#9
     writef(s);
 }
 void main(){
     f("mystery");
 }

 </code>

 The program above does not display "mystery" as expected.
 The string can be displayed when line #2 is removed
 or when line #9 is uncommented, both seemingly irrelevant.

 I've smelt something fishy in dmd's code generator.

 
 
 Can you reproduce this bug without using Phobo's functions?
 
 Thomas
 

I can reproduce the bug without calling std.utf.encode:
<code>
import std.stdio;
const char[] culprit;//#2
void f(...){
     char[] s;
     void putc(dchar c){
	s ~= c;		
     }
     std.format.doFormat(&putc, _arguments, _argptr);
     //foreach(char ch; s){} //#9
     writef(s);
}
void main(){
     f("mystery");
}
</code>

Dave <Dave_member pathlink.com> writes:

In article <d7tlh1$j0c$1 digitaldaemon.com>, zwang says...
Thomas Kühne wrote:
 
 Can you reproduce this bug without using Phobo's functions?
 
 Thomas
 

I can reproduce the bug without calling std.utf.encode:
<code>
import std.stdio;
const char[] culprit;//#2
void f(...){
     char[] s;
     void putc(dchar c){
	s ~= c;		
     }
     std.format.doFormat(&putc, _arguments, _argptr);
     //foreach(char ch; s){} //#9
     writef(s);
}
void main(){
     f("mystery");
}
</code>

Also, compiling w/ -O gives the expected output (both Linux and Win32).

- Dave

=?ISO-8859-1?Q?Thomas_K=FChne?= <thomas-dloop kuehne.THISISSPAM.cn> writes:

zwang wrote:

 <code>
 import std.stdio;
 const char[] culprit;//#2
 void f(...){
     char[] s;
     void putc(dchar c){
         std.utf.encode(s, c);
     }
     std.format.doFormat(&putc, _arguments, _argptr);
     //foreach(char ch; s){} //#9
     writef(s);
 }
 void main(){
     f("mystery");
 }

 </code>

 The program above does not display "mystery" as expected.
 The string can be displayed when line #2 is removed
 or when line #9 is uncommented, both seemingly irrelevant.

 I've smelt something fishy in dmd's code generator.

Added to DStress as
http://dstress.kuehne.cn/run/o/odd_bug_03_A.d
http://dstress.kuehne.cn/run/o/odd_bug_03_B.d
http://dstress.kuehne.cn/run/o/odd_bug_03_C.d

Thomas

"Walter" <newshound digitalmars.com> writes:

"zwang" <nehzgnaw gmail.com> wrote in message
news:d7d2kq$1mc$1 digitaldaemon.com...
 The program above does not display "mystery" as expected.

It's already been reported and the fix will go out in the next update.

zwang <nehzgnaw gmail.com> writes:

Walter wrote:
 "zwang" <nehzgnaw gmail.com> wrote in message
 news:d7d2kq$1mc$1 digitaldaemon.com...
 
The program above does not display "mystery" as expected.

 
 
 It's already been reported and the fix will go out in the next update.
 
 

Great! I'm wondering where the bug lurks.