I do not get a confirmation email from bugzilla after entering my e-mail address here: https://bugzilla.gdcproject.org/createaccount.cgi

I tried to build patched versions of GCC 4.8.5 and 4.9.4 but this happened:

.../gcc-4.8.5/libphobos/libdruntime/core/cpuid.d: In function 'cpuidX86':
.../gcc-4.8.5/libphobos/libdruntime/core/cpuid.d:838:17: error: PIC register clobbered by 'ebx' in ' asm'
"cpuid" : "=a" a, "=c" c : "a" 0x8000_0008 : "ebx", "edx";
^

patch branch gdc-4.8 commit 8b43f71621673c4d0f3f0665c397a1d80d5b434e

and

.../gcc-4.9.4/libphobos/libdruntime/core/cpuid.d: In function 'cpuidX86':
.../gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:838:17: error: PIC register clobbered by 'ebx' in ' asm'
"cpuid" : "=a" a, "=c" c : "a" 0x8000_0008 : "ebx", "edx";
^

patch branch gdc-4.9 commit 93a9105be173472a9e57aa12c43528ae15d1319c

Am Mon, 19 Feb 2018 19:43:49 +0000
schrieb kdevel <kdevel@vogtner.de>:

> I do not get a confirmation email from bugzilla after entering my e-mail address here: https://bugzilla.gdcproject.org/createaccount.cgi
> 
> I tried to build patched versions of GCC 4.8.5 and 4.9.4 but this happened:
> 
> .../gcc-4.8.5/libphobos/libdruntime/core/cpuid.d: In function
> 'cpuidX86':
> .../gcc-4.8.5/libphobos/libdruntime/core/cpuid.d:838:17: error:
> PIC register clobbered by 'ebx' in ' asm'
> "cpuid" : "=a" a, "=c" c : "a" 0x8000_0008 : "ebx", "edx";
> ^
> 
> patch branch gdc-4.8 commit 8b43f71621673c4d0f3f0665c397a1d80d5b434e
> 
> and
> 
> .../gcc-4.9.4/libphobos/libdruntime/core/cpuid.d: In function
> 'cpuidX86':
> .../gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:838:17: error:
> PIC register clobbered by 'ebx' in ' asm'
> "cpuid" : "=a" a, "=c" c : "a" 0x8000_0008 : "ebx", "edx";
> ^
> 
> patch branch gdc-4.9 commit 93a9105be173472a9e57aa12c43528ae15d1319c

Thanks for the report. That's a bug which was for some reason not caught by the CI. I guess it's a i686 only bug and IIRC that isn't tested right now.

I'll submit a fix later today or tomorrow.


-- Johannes

On Wednesday, 21 February 2018 at 07:58:29 UTC, Johannes Pfau wrote:
> Am Mon, 19 Feb 2018 19:43:49 +0000
> schrieb kdevel <kdevel@vogtner.de>:

[...]

> I'll submit a fix later today or tomorrow.

Great. But one question: Why did I receive an "noreply" notification email to my e-Mail-Address?

From the header:

From: D Programming Language Forum <no-reply@forum.dlang.org>
Return-Path: <no-reply@forum.dlang.org>
Received: from k3.1azy.net (k3.1azy.net [178.33.224.37])
	(using TLSv1.2 with cipher AES128-GCM-SHA256 (128/128 bits))
Subject: Johannes Pfau replied to your post in the thread "bugzilla.gdcproject.org: No confirmation mail; cpuid.d:838:17: error: PIC register clobbered by 'ebx' in ' asm'"

Am Wed, 21 Feb 2018 21:31:41 +0000
schrieb kdevel <kdevel%%@vogtner.de>:

> On Wednesday, 21 February 2018 at 07:58:29 UTC, Johannes Pfau wrote:
> > Am Mon, 19 Feb 2018 19:43:49 +0000
> > schrieb kdevel <kdevel@vogtner.de>:
> 
> [...]
> 
> > I'll submit a fix later today or tomorrow.
> 

I've pushed an update to the gdc-4.9 branch. Please report back whether this fixes the issue.

> Great. But one question: Why did I receive an "noreply" notification email to my e-Mail-Address?

I think this has been answered on the D newsgroup.

-- Johannes

On Thursday, 22 February 2018 at 19:52:12 UTC, Johannes Pfau wrote:
> I've pushed an update to the gdc-4.9 branch. Please report back whether this fixes the issue.

Not really:

checking for expf... /scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:833:45: error: basic type expected, not {
             version(GNU) pure nothrow @nogc {
                                             ^
/scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:833:45: error: no identifier for declarator _error_
             version(GNU) pure nothrow @nogc {
                                             ^
/scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:835:15: error: found 'else' instead of statement
             } else asm pure nothrow @nogc {
               ^
/scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:844:5: error: declaration expected, not 'if'
     if (max_extended_cpuid >= 0x8000_0004) {
     ^
/scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:848:21: error: unexpected ( in declarator
             gnuCpuid(0x8000_0002, pnb[0], pnb[1], pnb[ 2], pnb[ 3]);
                     ^
/scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:848:22: error: basic type expected, not 2147483650U
             gnuCpuid(0x8000_0002, pnb[0], pnb[1], pnb[ 2], pnb[ 3]);
                      ^
/scratch/mockbuild1/bld-4.9.4/gcc-4.9.4/libphobos/libdruntime/core/cpuid.d:848:22: error: found '2147483650U' when expecting ')'
             gnuCpuid(0x8000_0002, pnb[0], pnb[1], pnb[ 2], pnb[ 3]);

...

Am Thu, 22 Feb 2018 21:34:34 +0000
schrieb kdevel <kdevel@vogtner.de>:

> On Thursday, 22 February 2018 at 19:52:12 UTC, Johannes Pfau wrote:
> > I've pushed an update to the gdc-4.9 branch. Please report back whether this fixes the issue.
> 
> Not really:
> 

Looks like you're at commit c8ccdcfd1abf15b95b64821d5173e79f76c9dafe,
which is a broken version I accidentally committed.
The next commit, 8a6b7a49ff9d4233e637a6a7e1ca4c1978540060, should fix
this problem.


-- Johannes

On Friday, 23 February 2018 at 07:48:47 UTC, Johannes Pfau wrote:
> Am Thu, 22 Feb 2018 21:34:34 +0000
> schrieb kdevel <kdevel@vogtner.de>:
>
>> On Thursday, 22 February 2018 at 19:52:12 UTC, Johannes Pfau wrote:
>> > I've pushed an update to the gdc-4.9 branch. Please report back whether this fixes the issue.
>> 
>> Not really:
>> 
>
> Looks like you're at commit c8ccdcfd1abf15b95b64821d5173e79f76c9dafe,
> which is a broken version I accidentally committed.
> The next commit, 8a6b7a49ff9d4233e637a6a7e1ca4c1978540060, should fix
> this problem.

It does. Now I have an issue with a language feature. With dmd v2.078.2 this
code compiles:

void function () [string] disp;
static this () {
   enum typenames = ["float", "double", "real", "short", "int", "long"];
   static foreach (s; typenames) {
      pragma (msg, s);
      disp[s] = mixin ("&mymain!" ~ s);
   }
}

but gdc (GCC 4.9.4) complains:

dotbench.d:65:11: Fehler: basic type expected, not foreach
    static foreach (s; typenames) {
           ^
dotbench.d:65:11: Fehler: no identifier for declarator _error_
    static foreach (s; typenames) {
           ^

Am Sun, 25 Feb 2018 00:05:44 +0000 schrieb kdevel:
> 
> but gdc (GCC 4.9.4) complains:
> 
> dotbench.d:65:11: Fehler: basic type expected, not foreach
>      static foreach (s; typenames) {
>             ^
> dotbench.d:65:11: Fehler: no identifier for declarator _error_
>      static foreach (s; typenames) {
>             ^

Unfortunately, static foreach is not supported in GDC right now. Maybe we'll get a static foreach backport soon (https://dlang.slack.com/ archives/C6LTP6MV1/p1519419332000480 ), otherwise static foreach will only start working once we switch GDC to the D frontend (https:// github.com/D-Programming-GDC/GDC/pull/550 ).

-- 
Johannes

On Sunday, 25 February 2018 at 09:36:51 UTC, Johannes Pfau wrote:
> Am Sun, 25 Feb 2018 00:05:44 +0000 schrieb kdevel:
>> 
>> but gdc (GCC 4.9.4) complains:
>> 
>> dotbench.d:65:11: Fehler: basic type expected, not foreach
>>      static foreach (s; typenames) {
>>             ^
>> dotbench.d:65:11: Fehler: no identifier for declarator _error_
>>      static foreach (s; typenames) {
>>             ^
>
> Unfortunately, static foreach is not supported in GDC right now. Maybe we'll get a static foreach backport soon (https://dlang.slack.com/ archives/C6LTP6MV1/p1519419332000480 ), otherwise static foreach will only start working once we switch GDC to the D frontend (https:// github.com/D-Programming-GDC/GDC/pull/550 ).

I didn't come that far with the backport. Some bugs are very difficult to find (for me) and the D implementation uses some template features that aren't really translatable into pre-C++11 C++. So I'm not sure that I'll finish this. Especially I don't know if it is worth it. It depends how soon gdc will be merged with the upstream.