Something weird is going on then because:

1.  The error persists after I nuke my entire DMD installation and reinstall from scratch.

2.  The following ancient date stamps are on the linux/lib binaries extracted from the current zip file:

-rw-r--r--  1 dsimcha dsimcha   85 Feb  9  2009 README.TXT -rw-r--r--  1 dsimcha dsimcha 2.1M Mar 31  2009 libdruntime.a -rw-r--r--  1 dsimcha dsimcha 7.0M Sep 15 00:24 libphobos2.a


On Wed, Oct 27, 2010 at 5:17 PM, Walter Bright <walter at digitalmars.com>wrote:

>
>
> David Simcha wrote:
>
>> Is anyone besides me having trouble on Linux with  undefined reference to `_d_throwc' error messages?
>>
>>
> If that is happening, then you are not linking with the current libphobos.a.
>
> If you are linking with Tango, Tango needs to add that function.
>
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/dmd-beta/attachments/20101027/bd37054d/attachment.html>

Oh, I see the problem. The build directories changed, and so my install script broke. The same issue is on OSX. A fixed zip is uploaded. -Walter

David Simcha wrote:
> Something weird is going on then because:
>
> 1.  The error persists after I nuke my entire DMD installation and reinstall from scratch.
>
> 2.  The following ancient date stamps are on the linux/lib binaries extracted from the current zip file:
>
> -rw-r--r--  1 dsimcha dsimcha   85 Feb  9  2009 README.TXT -rw-r--r--  1 dsimcha dsimcha 2.1M Mar 31  2009 libdruntime.a -rw-r--r--  1 dsimcha dsimcha 7.0M Sep 15 00:24 libphobos2.a
>
>
> On Wed, Oct 27, 2010 at 5:17 PM, Walter Bright <walter at digitalmars.com <mailto:walter at digitalmars.com>> wrote:
>
>
>
>     David Simcha wrote:
>
>         Is anyone besides me having trouble on Linux with  undefined
>         reference to `_d_throwc' error messages?
>
>
>     If that is happening, then you are not linking with the current
>     libphobos.a.
>
>     If you are linking with Tango, Tango needs to add that function.
>
>     _______________________________________________
>     dmd-beta mailing list
>     dmd-beta at puremagic.com <mailto:dmd-beta at puremagic.com>
>     http://lists.puremagic.com/mailman/listinfo/dmd-beta
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta

Trying the old version of std.typecons and it works, so it's not a compiler bug.

Stephan Dilly wrote:
> I am sorry but i found another regression trying to build my codebase with the new beta:
>
>
> import std.typecons;
>
> alias Tuple!(int,"aa",int,"bb") Foo;
> 
> void main()
> {
>     Foo[] m_array;
>
>     Foo f;
> 
>     m_array ~= f;
> }
>
>
> This portion of code used to work with dmd2049. The beta now says:
> Error: cannot append type (int,int) to type Tuple!(int,"aa",int,"bb")[]
>

Then it is a regression in phobos, which doesn't make it less of a regression that should be fixed b4 release (imho).


On 28.10.2010 00:51, Walter Bright wrote:
> Trying the old version of std.typecons and it works, so it's not a compiler bug.
>
> Stephan Dilly wrote:
>> I am sorry but i found another regression trying to build my codebase with the new beta:
>>
>>
>> import std.typecons;
>>
>> alias Tuple!(int,"aa",int,"bb") Foo;
>>     void main()
>> {
>>     Foo[] m_array;
>>
>>     Foo f;
>>            m_array ~= f;
>> }
>>
>>
>> This portion of code used to work with dmd2049. The beta now says:
>> Error: cannot append type (int,int) to type Tuple!(int,"aa",int,"bb")[]
>>
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta
>

The problem is caused by line 362 in typecons.d, introduced by Shin (rsinfu):

alias field this;

Commenting that line out makes Stephan's code sample work.

Shin's change is beneficial because it allows using tup[2] to access the third field of a tuple "tup". This was one of the main intents of the "alias this" feature. Using the workarounds such as tup.field[2] or tup._2 has never gained traction, so I'm glad to see top[2] (almost) work.

The change does not introduce a bug, instead it exposes a bug in the compiler. The bug is that the compiler is too eager to automatically trigger the "alias this" mechanism. I therefore suggest making Stephan's code into a unittest and fixing the compiler to make the unittest work.


Andrei

On 10/27/10 17:51 CDT, Walter Bright wrote:
> Trying the old version of std.typecons and it works, so it's not a compiler bug.
>
> Stephan Dilly wrote:
>> I am sorry but i found another regression trying to build my codebase with the new beta:
>>
>>
>> import std.typecons;
>>
>> alias Tuple!(int,"aa",int,"bb") Foo;
>> void main()
>> {
>> Foo[] m_array;
>>
>> Foo f;
>> m_array ~= f;
>> }
>>
>>
>> This portion of code used to work with dmd2049. The beta now says:
>> Error: cannot append type (int,int) to type Tuple!(int,"aa",int,"bb")[]
>>
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta

Stephan Dilly wrote:
>
> This portion of code used to work with dmd2049. The beta now says:
> Error: cannot append type (int,int) to type Tuple!(int,"aa",int,"bb")[]
>
>

This is fixed now, new beta uploaded.

Yup all my code seems to work now again. thanks for the effort.

Regards,
	Stephan


On 29.10.2010 03:56, Walter Bright wrote:
>
>
> Stephan Dilly wrote:
>>
>> This portion of code used to work with dmd2049. The beta now says:
>> Error: cannot append type (int,int) to type Tuple!(int,"aa",int,"bb")[]
>>
>>
>
> This is fixed now, new beta uploaded.
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta
>

Stephan Dilly wrote:
>
> Yup all my code seems to work now again. thanks for the effort.
>
> Regards,
>     Stephan
>

You're welcome. If nothing else comes up by tomorrow, I plan on doing a release.

as nobody managed to reduce the bug I had found compiling tango, I did
it now, it is
	http://d.puremagic.com/issues/show_bug.cgi?id=5145
it is a regression connected to a forward ref.

Fawzi

Fawzi Mohamed wrote:
> as nobody managed to reduce the bug I had found compiling tango, I did
> it now, it is
>     http://d.puremagic.com/issues/show_bug.cgi?id=5145
> it is a regression connected to a forward ref.
>

Thanks!