Glad to announce the first beta for the 2.085.0 release, ♥ to the 49 contributors.

http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.085.0.html

As usual please report any bugs at
https://issues.dlang.org

-Martin

On Saturday, 16 February 2019 at 15:06:51 UTC, Martin Nowak wrote:
> Glad to announce the first beta for the 2.085.0 release, ♥ to the 49 contributors.
>
> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.085.0.html
>
> As usual please report any bugs at
> https://issues.dlang.org
>
> -Martin

What is the precise scanning GC? I only see how to enable it, but don't really see why would I want to enable it.

-verrors=context

Why this is not default behavior?

On Saturday, 16 February 2019 at 19:21:12 UTC, Suliman wrote:
> -verrors=context
>
> Why this is not default behavior?

It's the first release with them and the implementation may not be perfect (e.g. mixins and other corner cases).

On Saturday, 16 February 2019 at 19:21:12 UTC, Suliman wrote:
> -verrors=context
>
> Why this is not default behavior?

My guess is that it'd break the IDEs which parse DMD output for error messages.

On Saturday, 16 February 2019 at 15:06:51 UTC, Martin Nowak wrote:
> Glad to announce the first beta for the 2.085.0 release, ♥ to the 49 contributors.
>
> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.085.0.html

I can't find any mentioning of Rainer Schützes work on memory reductions of the GC for non-power-of-two sizes. Was it reverted?

Thanks for all the work!

Thank you, Martin!

On 16/02/2019 19:53, JN wrote:
> On Saturday, 16 February 2019 at 15:06:51 UTC, Martin Nowak wrote:
>> Glad to announce the first beta for the 2.085.0 release, ♥ to the 49 contributors.
>>
>> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.085.0.html
>>
>> As usual please report any bugs at
>> https://issues.dlang.org
>>
>> -Martin
> 
> What is the precise scanning GC? I only see how to enable it, but don't really see why would I want to enable it.

It's purpose is to avoid not collecting memory due to false pointers. Here are some related bug reports:

https://issues.dlang.org/show_bug.cgi?id=3463 https://issues.dlang.org/show_bug.cgi?id=4358 https://issues.dlang.org/show_bug.cgi?id=9094 https://issues.dlang.org/show_bug.cgi?id=13801 https://issues.dlang.org/show_bug.cgi?id=18900

Precise heap and data segment scanning isn't the complete solution, though, The stack and registers might still contain false pointers.

On 16/02/2019 22:03, Per Nordlöw wrote:
> On Saturday, 16 February 2019 at 15:06:51 UTC, Martin Nowak wrote:
>> Glad to announce the first beta for the 2.085.0 release, ♥ to the 49 contributors.
>>
>> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.085.0.html
> 
> I can't find any mentioning of Rainer Schützes work on memory reductions of the GC for non-power-of-two sizes. Was it reverted?

It is listed as "Bugzilla 19455: GC wastes too much memory".

What's not found in the changelog is an optimization sweeping large allocations that is inspired by your report https://forum.dlang.org/thread/zrouxhbkssrncyinhvdz@forum.dlang.org: https://github.com/dlang/druntime/pull/2430

On Saturday, 16 February 2019 at 15:06:51 UTC, Martin Nowak wrote:
> Glad to announce the first beta for the 2.085.0 release, ♥ to the 49 contributors.
>
> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.085.0.html
>
> As usual please report any bugs at
> https://issues.dlang.org
>
> -Martin

-transition=intpromote leads to compile error instead of continue to work as it stated in the docs.