On 2015-10-14 20:32, John Colvin wrote:

> got through to homebrew a faster this time.
>
> brew reinstall dmd --devel

It's always available through DVM 1 second after announcement :)

-- 
/Jacob Carlborg

On Wednesday, 14 October 2015 at 13:53:17 UTC, Martin Nowak wrote:
> Second beta for the 2.069.0 release.
>
> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.069.0.html
>
> Please report any bugs at https://issues.dlang.org
>
> -Martin

Apart from https://issues.dlang.org/show_bug.cgi?id=15207, I'm seeing a huge +170% speed-up in 32-bit mode for optimized builds vs 2.068, something that is well appreciated :) 64-bit performance is mostly the same.
What changed in the backend?

On Friday, 16 October 2015 at 12:14:09 UTC, ponce wrote:
> What changed in the backend?

http://dlang.org/changelog/2.069.0.html#backend-improvements

On Saturday, 17 October 2015 at 08:13:03 UTC, Martin Nowak wrote:
> On Friday, 16 October 2015 at 12:14:09 UTC, ponce wrote:
>> What changed in the backend?
>
> http://dlang.org/changelog/2.069.0.html#backend-improvements

Is that limited to "-release" builds ?

Am Wed, 14 Oct 2015 15:52:57 +0200
schrieb Martin Nowak <code+news.digitalmars@dawg.eu>:

> Second beta for the 2.069.0 release.
> 
> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.069.0.html
> 
> Please report any bugs at https://issues.dlang.org
> 
> -Martin

When I use a specific host compiler, it still picks up the dmd.conf provided in the package and doesn't find object.d. Should I manually delete dmd.conf before building?

-- 
Marco

On Friday, 16 October 2015 at 12:14:09 UTC, ponce wrote:
> On Wednesday, 14 October 2015 at 13:53:17 UTC, Martin Nowak wrote:
>> Second beta for the 2.069.0 release.
>>
>> http://dlang.org/download.html#dmd_beta http://dlang.org/changelog/2.069.0.html
>>
>> Please report any bugs at https://issues.dlang.org
>>
>> -Martin
>
> Apart from https://issues.dlang.org/show_bug.cgi?id=15207, I'm seeing a huge +170% speed-up in 32-bit mode for optimized builds vs 2.068, something that is well appreciated :) 64-bit performance is mostly the same.
> What changed in the backend?

When I wrote the section on backend improvements for the changelog, I wasn't aware how much faster the 32 bit code was. Would you be willing to run any benchmarks so I can add in actual numbers to that section.

On Saturday, 17 October 2015 at 09:24:42 UTC, extrawurst wrote:
> On Saturday, 17 October 2015 at 08:13:03 UTC, Martin Nowak wrote:
>> On Friday, 16 October 2015 at 12:14:09 UTC, ponce wrote:
>>> What changed in the backend?
>>
>> http://dlang.org/changelog/2.069.0.html#backend-improvements
>
> Is that limited to "-release" builds ?

No, they're limited to -O builds.

Oh wait, false alert. That was a relic from older days. My build script placed a dummy dmd.conf there.

I do seem to get problems with ldc2-0.16.0:

  make -C druntime -f posix.mak MODEL=32
  ../dmd/src/dmd -conf= -c -o- -Isrc -Iimport -Hfimport/core/sync/barrier.di src/core/sync/barrier.d
  core.exception.AssertError@expression.d(4369): Assertion failure

That is this line of code: https://github.com/D-Programming-Language/dmd/blob/v2.069.0-b2/src/expression.d#L4369

Stack trace (with file + line numbers now, hey!):
#0  StringExp::compare(RootObject*) (this=0xb66e30, obj=0xb65c80) at expression.d:4341
#1  0x00000000004fb6ed in StringExp::equals(RootObject*) (this=0xb66e30, o=0xb65c80) at expression.d:4175
#2  0x00000000004c4fe9 in match(RootObject*, RootObject*) (o1=0xb66e30, o2=0xb65c80) at dtemplate.d:246
#3  0x00000000004c51c6 in arrayObjectMatch(Array<RootObject*>*, Array<RootObject*>*) (oa1=0x7ffff64aca98, oa2=0x7ffff64ac898) at dtemplate.d:290
#4  0x00000000004cddd7 in TemplateInstance::compare(RootObject*) (this=0x7ffff64aca00, o=0x7ffff64ac800) at dtemplate.d:6232
#5  0x00000000004cdaf8 in TemplateDeclaration::findExistingInstance(TemplateInstance*, Array<Expression*>*) (this=0x7ffff64ac600, tithis=0x7ffff64aca00, fargs=0x0) at dtemplate.d:2039
#6  0x00000000004d2f24 in TemplateInstance::semantic(Scope*, Array<Expression*>*) (this=0x7ffff64aca00, sc=0x7ffff65dfc00, fargs=0x0) at dtemplate.d:5583
#7  0x0000000000406877 in TemplateInstance::semantic(Scope*) (this=0x7ffff64aca00, sc=0x7ffff65dfc00) at dtemplate.d:5967
#8  0x000000000057a03a in TypeInstance::resolve(Loc, Scope*, Expression**, Type**, Dsymbol**, bool) (this=0x7ffff64ae100, loc=..., sc=0x7ffff65dfc00, pe=0x7fffffffcec0, pt=0x7fffffffcec8, ps=0x7fffffffceb8, intypeid=false) at mtype.d:7412
#9  0x000000000057a327 in TypeInstance::toDsymbol(Scope*) (this=0x7ffff64ae100, sc=0x7ffff65dfc00) at mtype.d:7459
#10 0x000000000043c5d6 in AliasDeclaration::semantic(Scope*) (this=0x7ffff64ae200, sc=0x7ffff65dfc00) at .:598
#11 0x00000000004897f9 in Module::semantic() (this=0x7ffff6378400) at dmodule.d:976
#12 0x0000000000488e0f in Import::semantic(Scope*) (this=0x7ffff6a82800, sc=0x7ffff6aa9c00) at dimport.d:258
#13 0x000000000042759a in AttribDeclaration::semantic(Scope*) (this=0x7ffff6a82900, sc=0x7ffff6aa9600) at attrib.d:168
#14 0x00000000004897f9 in Module::semantic() (this=0x7ffff6a7fe00) at dmodule.d:976
#15 0x0000000000488e0f in Import::semantic(Scope*) (this=0x7ffff7eca100, sc=0x7ffff7ed2200) at dimport.d:258
#16 0x000000000042759a in AttribDeclaration::semantic(Scope*) (this=0x7ffff7eca200, sc=0x7ffff7ecee00) at attrib.d:168
#17 0x00000000004897f9 in Module::semantic() (this=0x7ffff7ec9000) at dmodule.d:976
#18 0x0000000000567de5 in tryMain(unsigned long, char const**) (argc=8, argv=0x7fffffffe978) at mars.d:1485
#19 0x000000000056a567 in D main () at mars.d:1695

`sz` is 0, `string` and `len` seem to be ok.

-- 
Marco

On 10/17/2015 09:05 PM, Marco Leise wrote:
> Oh wait, false alert. That was a relic from older days. My build script placed a dummy dmd.conf there.
> 
> I do seem to get problems with ldc2-0.16.0:

Are you using something befor 0.16.0-beta2, b/c I thought the problem
was resolved.
https://github.com/D-Programming-Language/dmd/pull/5025#issuecomment-142143727

Am Tue, 20 Oct 2015 19:26:13 +0200
schrieb Martin Nowak <code+news.digitalmars@dawg.eu>:

> On 10/17/2015 09:05 PM, Marco Leise wrote:
> > Oh wait, false alert. That was a relic from older days. My build script placed a dummy dmd.conf there.
> > 
> > I do seem to get problems with ldc2-0.16.0:
> 
> Are you using something befor 0.16.0-beta2, b/c I thought the problem
> was resolved.
> https://github.com/D-Programming-Language/dmd/pull/5025#issuecomment-142143727

Indeed I should have checked that. I'm using 0.16.0_alpha4. Alright then. Everything works as designed now. :)

-- 
Marco