Hi all!

Yesterday, the new DMD release 2.067.0 was announced.

I am glad to tell you that the merge-2.067 branch is now at the release level, too. Thanks to the help from kinke and smolt the compiler is already quite usable. Well, we still face some serious issues.

If you like to help please have a look at issue #868 and check the remaining test failures in the dmd testsuite!

Regards,
Kai

On Wednesday, 25 March 2015 at 05:49:34 UTC, Kai Nacke wrote:
> Hi all!
>
> Yesterday, the new DMD release 2.067.0 was announced.
>
> I am glad to tell you that the merge-2.067 branch is now at the release level, too. Thanks to the help from kinke and smolt the compiler is already quite usable. Well, we still face some serious issues.
>
> If you like to help please have a look at issue #868 and check the remaining test failures in the dmd testsuite!
>
> Regards,
> Kai

Wow, that was fast...I always thought that ldc was at least one step behind dmd frontend spec.

On Wednesday, 25 March 2015 at 18:30:56 UTC, Baz wrote:
> Wow, that was fast...I always thought that ldc was at least one step behind dmd frontend spec.

Hi Baz!

You should not expect a release very soon - there are still serious bugs to fix. But at least the merge is done. Even that took some time in the past.

(I try to follow DMD more closely. I started the merge even before the first beta release.)

Regards,
Kai

Hi Kai !
I wonder why it's a problem to follow dmd's frontend.
90% of pulls is frontend bug fixes. I think there are no so much
changes with interacting with backend ?

On 03/25/2015 09:22 PM, Temtaime via digitalmars-d-ldc wrote:
> Hi Kai !
> I wonder why it's a problem to follow dmd's frontend.
> 90% of pulls is frontend bug fixes. I think there are no so much
> changes with interacting with backend ?

This would be true if the AST generated by the frontend was static. However, its exact form is unfortunately quite fluid. Some of the problematic frontend changes in fact fix a quirk that we needed to work around, some of them introduce new bugs that happen not to matter for the way the DMD glue layer uses the AST, and some just change the AST structure (e.g. by moving something out of the glue layer into the frontend or vice versa).

 — David

Kai Nacke:

> Hi all!

See also my post here:

http://forum.dlang.org/thread/liqxeuzwyqzgfgcitefp@forum.dlang.org#post-rzsrvmnjhkalbjpdaibp:40forum.dlang.org

Bye,
bearophile

Hi Kay !
Great work. I can build hello world on win64.
Will test further.

Hi again Kai !
I'm wondering if --disable-linker-strip-dead flag works because when i enable it link.exe produces many of errors like:

phobos64.lib(bits.obj) : error LNK2005: _D11TypeInfo_Pv6__initZ already defined in main.obj
phobos64.lib(demangle.obj) : error LNK2005: _D12TypeInfo_xAa6__initZ already defined in main.obj
phobos64.lib(demangle.obj) : error LNK2005: _D11TypeInfo_xm6__initZ already defined in main.obj

...

sorry my fault

On Saturday, 9 May 2015 at 13:16:43 UTC, Temtaime wrote:
> Hi Kay !
> Great work. I can build hello world on win64.
> Will test further.

The thanks goes to Martin, Johan and Dan for their great work on the 2.067 merge and the Win64 version!

Regards,
Kai