On 23 October 2011 07:29, Don Clugston <dclugston at googlemail.com> wrote:
> On 22 October 2011 23:56, Walter Bright <walter at digitalmars.com> wrote:
>> http://ftp.digitalmars.com/dmd1beta.zip http://ftp.digitalmars.com/dmd2beta.zip
>
> There are still two open regressions relative to 2.055:
> http://d.puremagic.com/issues/show_bug.cgi?id=6825 (which has a pull request)
> http://d.puremagic.com/issues/show_bug.cgi?id=6753 (bisected, looks
> fairly simple)

The increase in compile time, on Windows at least, is also a serious problem. Even "Hello World" takes about twice as long to compile, presumably because of the new compiler memory management. This might be tolerable if memory usage bugs like bug 1382 were fixed, but 1382 still fails. So it just seems to be a completely backwards step. So I really think the new memory management should be disabled until the effect of compilation is less drastic.

On 10/23/2011 12:02 PM, Don Clugston wrote:
>
> The increase in compile time, on Windows at least, is also a serious problem. Even "Hello World" takes about twice as long to compile, presumably because of the new compiler memory management. This might be tolerable if memory usage bugs like bug 1382 were fixed, but 1382 still fails. So it just seems to be a completely backwards step. So I really think the new memory management should be disabled until the effect of compilation is less drastic.
>

I agree. I'll yank it.

On 10/22/2011 2:56 PM, Walter Bright wrote:
> http://ftp.digitalmars.com/dmd1beta.zip http://ftp.digitalmars.com/dmd2beta.zip

Only 5 more bugs left to have a "no new known regressions" release:

6669 Compiler seg fault when using square brackets in inline assembly
  -- might be fixed.. osx specific issue that isn't reproducable with the recent beta

6681 bogus duplicate union initialization or overlapping initialization errors

6698 Regression(2.053): segfault with naked asm in inner function
  -- not actually a recent regression, test case was broken for a long time

6700 Regression(2.053) using $ inside a slice of a tuple
  -- not actually a recent regression, test case was broken for a long time

6744 Missing AA symbol with enum

Though not a regression, I'd like this to be fixed:

6096 optimizer assert on cdouble to bool conversion (fix: https://github.com/D-Programming-Language/dmd/pull/452)

It is keeping visual d from building with the stock compiler.

On 23.10.2011 22:00, Brad Roberts wrote:
> On 10/22/2011 2:56 PM, Walter Bright wrote:
>> http://ftp.digitalmars.com/dmd1beta.zip http://ftp.digitalmars.com/dmd2beta.zip
> Only 5 more bugs left to have a "no new known regressions" release:
>
> 6669 Compiler seg fault when using square brackets in inline assembly
>    -- might be fixed.. osx specific issue that isn't reproducable with the recent beta
>
> 6681 bogus duplicate union initialization or overlapping initialization errors
>
> 6698 Regression(2.053): segfault with naked asm in inner function
>    -- not actually a recent regression, test case was broken for a long time
>
> 6700 Regression(2.053) using $ inside a slice of a tuple
>    -- not actually a recent regression, test case was broken for a long time
>
> 6744 Missing AA symbol with enum
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta
>

On 23.10.2011 21:42, Walter Bright wrote:
>
>
> On 10/23/2011 12:02 PM, Don Clugston wrote:
>>
>> The increase in compile time, on Windows at least, is also a serious problem. Even "Hello World" takes about twice as long to compile, presumably because of the new compiler memory management. This might be tolerable if memory usage bugs like bug 1382 were fixed, but 1382 still fails. So it just seems to be a completely backwards step. So I really think the new memory management should be disabled until the effect of compilation is less drastic.
>>
>
> I agree. I'll yank it.

I always wonder why dmd runs out of memory at about 1GB. My guess is that it calls VirtualAlloc on smaller blocks than the allocation granularity (64kB).

A one-bit fix to get more memory for the win32 process is to set the LARGE_ADDRESS_AWARE bit in the file header of dmd.exe.

> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta
>

On 10/23/2011 07:54 PM, David Simcha wrote:
> Was there some change made to error checking for virtual functions/interfaces?  I just tried to compile gtkD with the beta and I got the error message "Ambiguous Virtual Function Call".  IDK if gtkD was relying on a DMD bug that was fixed or if this is a regression in DMD.  What does this error message mean?
>

Fixed: http://www.dsource.org/projects/gtkd/changeset/904
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/dmd-beta/attachments/20111024/85543d9e/attachment.html>

From: "Nick Sabalausky" <bus_dmdbeta at semitwist.com>
>I hope it's not too late, but it would be really nice if these two RDMD pull requests could be merged and included in the new release:
>
> https://github.com/D-Programming-Language/tools/pull/14 https://github.com/D-Programming-Language/tools/pull/15
>
> RDMD bugs such as these can be a real pain since they'll break your build process. The first one prevents RDMD from working with VisualD. The second one often prevents RDMD from working on projects that use static libs.
>
> Both bugs make D look bad to newcomers who often expect a toolchain to automatically find and compile dependencies...Such as here:
>
> http://www.linuxquestions.org/questions/programming-9/d-is-d-programming-language-just-too-much-898862/#post4452127
>

I take it that's a "no" then. That'll be unfortunate for those to have to wait a release cycle.

On 10/23/2011 10:05 PM, Nick Sabalausky wrote:
> From: "Nick Sabalausky" <bus_dmdbeta at semitwist.com>
>> I hope it's not too late, but it would be really nice if these two RDMD pull requests could be merged and included in the new release:
>>
>> https://github.com/D-Programming-Language/tools/pull/14 https://github.com/D-Programming-Language/tools/pull/15
>>
>> RDMD bugs such as these can be a real pain since they'll break your build process. The first one prevents RDMD from working with VisualD. The second one often prevents RDMD from working on projects that use static libs.
>>
>> Both bugs make D look bad to newcomers who often expect a toolchain to automatically find and compile dependencies...Such as here:
>>
>> http://www.linuxquestions.org/questions/programming-9/d-is-d-programming-language-just-too-much-898862/#post4452127
>>
>>
>
> I take it that's a "no" then. That'll be unfortunate for those to have to wait a release cycle.
>

I plan to pull them, just haven't gotten to it yet.

From: "Walter Bright" <walter at digitalmars.com>
> On 10/23/2011 10:05 PM, Nick Sabalausky wrote:
>> From: "Nick Sabalausky" <bus_dmdbeta at semitwist.com>
>>> I hope it's not too late, but it would be really nice if these two RDMD pull requests could be merged and included in the new release:
>>>
>>> https://github.com/D-Programming-Language/tools/pull/14 https://github.com/D-Programming-Language/tools/pull/15
>>>
>>> RDMD bugs such as these can be a real pain since they'll break your build process. The first one prevents RDMD from working with VisualD. The second one often prevents RDMD from working on projects that use static libs.
>>>
>>> Both bugs make D look bad to newcomers who often expect a toolchain to automatically find and compile dependencies...Such as here:
>>>
>>> http://www.linuxquestions.org/questions/programming-9/d-is-d-programming-language-just-too-much-898862/#post4452127
>>>
>>
>> I take it that's a "no" then. That'll be unfortunate for those to have to wait a release cycle.
>>
>
> I plan to pull them, just haven't gotten to it yet.

Fantastic :)

And then there was one.. just:

  http://d.puremagic.com/issues/show_bug.cgi?id=6681

Almost to the point of a 0 regression release.  Hopefully the start of a very important and wonderful trend.

Thanks everyone,
Brad

On Sun, 23 Oct 2011, Brad Roberts wrote:

> Only 5 more bugs left to have a "no new known regressions" release:
> 
> 6669 Compiler seg fault when using square brackets in inline assembly
>   -- might be fixed.. osx specific issue that isn't reproducable with the recent beta
> 
> 6681 bogus duplicate union initialization or overlapping initialization errors
> 
> 6698 Regression(2.053): segfault with naked asm in inner function
>   -- not actually a recent regression, test case was broken for a long time
> 
> 6700 Regression(2.053) using $ inside a slice of a tuple
>   -- not actually a recent regression, test case was broken for a long time
> 
> 6744 Missing AA symbol with enum