phobos commit, revision 1552


user: rsinfu

msg:
Fixed bugzilla 3465: isIdeographic can be wrong in std.xml.

- Applied Michael Rynn's contribution to isChar(), isDigit() and isIdeographic(). Now these functions are hard-coded for performance. Profiling showed that this code was three times faster than the old one.
- IdeographicTable must be sorted for lookup().

http://www.dsource.org/projects/phobos/changeset/1552

The unittests in std.xml throw an access violation.

dsource.org wrote:
> phobos commit, revision 1552
>
>
> user: rsinfu
>
> msg:
> Fixed bugzilla 3465: isIdeographic can be wrong in std.xml.
>
> - Applied Michael Rynn's contribution to isChar(), isDigit() and isIdeographic(). Now these functions are hard-coded for performance. Profiling showed that this code was three times faster than the old one.
> - IdeographicTable must be sorted for lookup().
>
> http://www.dsource.org/projects/phobos/changeset/1552
>
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>
>
> 

> The unittests in std.xml throw an access violation.
> 

I'm sorry but I can't reproduce it on FreeBSD, Gentoo Linux nor Wine.

I believe changeset 1552 does not contain anything that can cause access violation. The access violation was thrown by std.functional, wasn't it?


Shin

Shin Fujishiro wrote:
>> The unittests in std.xml throw an access violation.
>>
>> 
>
> I'm sorry but I can't reproduce it on FreeBSD, Gentoo Linux nor Wine.
>
> I believe changeset 1552 does not contain anything that can cause access violation. The access violation was thrown by std.functional, wasn't it?
>
>
> 

It's dying in Windows. The debugger says it was in std.xml, but it might be in error.

It segfaults for me too, but it's not std.xml, it's the previously reported issue with toDelegate, which still isn't fixed in trunk: Quoting shin:
-----

obj2asm showed that everything in toDelegate() was optimized away if compiled with the -O option. Then an uninitialized delegate object was returned.

Workaround:
--------------------
--- phobos/std/functional.d
+++ phobos/std/functional.d
@@ -590,7 +590,8 @@ auto ref toDelegate(F)(auto ref F fp) if
(isCallable!(F)) { auto dummyDel = &(dummy.doIt);
        df.funcPtr = dummyDel.funcptr;

-        return df.del;
+        auto del = df.del;
+        return del;
    }
 }

--------------------

toDelegate() is an auto ref function. A compiler bug allows the local
variable df.del to be returned by reference. I'm not very sure, but it
leads toDelegate() to return (reference to) uninitialized value.

I filed the bug into bugzilla: http://d.puremagic.com/issues/show_bug.cgi?id=4232



On 26 May 2010 12:46, Walter Bright <walter at digitalmars.com> wrote:
>
>
> Shin Fujishiro wrote:
>>>
>>> The unittests in std.xml throw an access violation.
>>>
>>>
>>
>> I'm sorry but I can't reproduce it on FreeBSD, Gentoo Linux nor Wine.
>>
>> I believe changeset 1552 does not contain anything that can cause access violation. The access violation was thrown by std.functional, wasn't it?
>>
>>
>>
>
> It's dying in Windows. The debugger says it was in std.xml, but it might be
> in error.
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>

On 05/26/2010 05:46 AM, Walter Bright wrote:
>
>
> Shin Fujishiro wrote:
>>> The unittests in std.xml throw an access violation.
>>>
>>
>> I'm sorry but I can't reproduce it on FreeBSD, Gentoo Linux nor Wine.
>>
>> I believe changeset 1552 does not contain anything that can cause access violation. The access violation was thrown by std.functional, wasn't it?
>>
>>
>
> It's dying in Windows. The debugger says it was in std.xml, but it might be in error.

Per Don, it's not std.xml.

My oh my. I seem to recall there was a guy who kept on saying crashes are not a problem! Debuggers take care of them! Stop worrying about the null pointer!

Who was that guy? I'd like to see him piping up now.


Andrei

On 26/05/10 14:27, Andrei Alexandrescu wrote:
> Per Don, it's not std.xml.
>
> My oh my. I seem to recall there was a guy who kept on saying crashes are not a problem! Debuggers take care of them! Stop worrying about the null pointer!
>
> Who was that guy? I'd like to see him piping up now.
>
>
> Andrei

*pipes up* I'd fire up a debugger and take a look, I don't have a windows install though. I stand by my statement, out of the hundreds of crashes I've dealt with a debugger (gdb) has always sorted it for me.

Andrei Alexandrescu wrote:
>
> Per Don, it's not std.xml.
>
> My oh my. I seem to recall there was a guy who kept on saying crashes are not a problem! Debuggers take care of them! Stop worrying about the null pointer!
>
> Who was that guy? I'd like to see him piping up now.
>

Gee, I spent a whole 2 minutes looking at the problem.

On Wed, May 26, 2010 at 11:09 PM, Walter Bright <walter at digitalmars.com> wrote:
>
> Gee, I spent a whole 2 minutes looking at the problem.

"Cure me doctor, I'm ill."
"Why bother? It's easier to take 2 pills every day to suppress the symptoms."

A problem which is easy to solve is still a problem.

An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/phobos/attachments/20100526/be804ebd/attachment.html>