On 6/1/13, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
> So it looks like an RDMD-related bug.

Ah it's definitely an RDMD bug, look:

$ rdmd --chatty --force --compiler=dmd_msc test.d

It then spits out these and just exits:

-----
stat C:\Users\ADMINI~1\AppData\Local\Temp\.rdmd
stat C:\Users\ADMINI~1\AppData\Local\Temp\.rdmd\rdmd-test.d-1648EF87D7393B7077F5EA9F7EC96F7F
"dmd_msc" "-v" "-o-" "test.d" "-I."
read C:\Users\ADMINI~1\AppData\Local\Temp\.rdmd\rdmd-test.d-1648EF87D7393B7077F5EA9F7EC96F7F\rdmd.deps
stat C:\Windows\System32\dmd_msc
stat F:\path\dmd_msc
stat C:\Program Files (x86)\Windows Resource Kits\Tools\dmd_msc
stat %SystemRoot%\system32\dmd_msc
stat %SystemRoot%\dmd_msc
stat %SystemRoot%\System32\Wbem\dmd_msc
stat %SYSTEMROOT%\System32\WindowsPowerShell\v1.0\dmd_msc
stat C:\Program Files\Microsoft Windows Performance Toolkit\dmd_msc
stat C:\Program Files (x86)\Gtk-Runtime\bin\dmd_msc
stat C:\Program Files (x86)\GtkSharp\2.12\bin\dmd_msc
stat C:\Program Files (x86)\dub\dmd_msc
stat C:\dmd-git\dmd2\src\dmd_msc
stat C:\Program Files (x86)\FLAC\dmd_msc
stat C:\dmd-git\dmd2\windows\bin\dmd_msc
stat C:\DMD\dmd2\windows\bin\dmd_msc
stat C:\dm\bin\dmd_msc
stat C:\MinGW\bin\dmd_msc
stat D:\Apps\Console2\dmd_msc
stat D:\Program Files\GnuWin32\bin\dmd_msc
stat C:\Program Files (x86)\IrfanView\dmd_msc
-----

But if I add an .exe extension, it compiles, and runs the unittests:

$ rdmd --chatty --force --compiler=dmd_msc.exe test.d

On 6/1/13, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
> Other than that it does seem to be faster. I haven't ran any benchmarks yet.

Now that I've fixed my builds, it does seem to run considerably faster:

DMD with DMC:
Done in 8_331_946 usecs.

DMD with MSC:
Done in 3_948_409 usecs.

This is more than twice as fast. Pretty damn amazing.

On 6/1/13, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
> On 6/1/13, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
>> So it looks like an RDMD-related bug.
>
> Ah it's definitely an RDMD bug.

Filed as http://d.puremagic.com/issues/show_bug.cgi?id=10229

On 01.06.2013 15:35, Andrej Mitrovic wrote:
> On 6/1/13, Benjamin Thaut <code@benjamin-thaut.de> wrote:
>> There is a bug in the visual studio 2010 and up compiler which causes
>> real support to break.
>
> Ouch. I'll give it a go soon. Is this filed as a bug / pull ready somewhere?
>

I've made a pull request:

https://github.com/D-Programming-Language/dmd/pull/2119

On Saturday, 1 June 2013 at 14:52:25 UTC, Rainer Schuetze wrote:
>> Benjamin Thaut
>
> If I remember correctly you also had problems with a MSVC build invoking the linker if there are spaces in the path to it. I suspect this might also happen when rdmd tries to run the program.


You are correct, I vaguley remember that issues. It was a issue
with the microsoft c-runtime. Should I do a pull request for that?

Right, so the original question was, would whoever does the dmd releases
consider building the release binaries with msvc? Is it Walter? Doubling
the compiler's performance is significant. CTFE heavy code gets a much
bigger boost too in my experience.
On 2 Jun 2013 01:03, "Andrej Mitrovic" <andrej.mitrovich@gmail.com> wrote:

> On 6/1/13, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
> > On 6/1/13, Andrej Mitrovic <andrej.mitrovich@gmail.com> wrote:
> >> So it looks like an RDMD-related bug.
> >
> > Ah it's definitely an RDMD bug.
>
> Filed as http://d.puremagic.com/issues/show_bug.cgi?id=10229
>

On Sunday, June 02, 2013 08:33:20 Manu wrote:
> Right, so the original question was, would whoever does the dmd releases consider building the release binaries with msvc? Is it Walter? Doubling the compiler's performance is significant. CTFE heavy code gets a much bigger boost too in my experience.

At this point, the whole release is done by Walter locally. He _might_ have now given enough information to reproduce it recently when some folks were bugging him about it, but it involves running a set of commands that are only known to him and anyone who read the posting where he gave them. We really need to get to the point that they're being generated by something like autotester (potentially on a nightly basis as well so that anyone can easily grab the latest code if they want to mess around with the current development version).

Andrei and others have been pointing out for a while that all of the testing and release stuff that Walter does locally should be automated and reproducible by others, and we're getting there bit by bit, but we aren't there quite yet.

- Jonathan M Davis

On Sun, 2 Jun 2013 08:33:20 +1000
Manu <turkeyman@gmail.com> wrote:

> Right, so the original question was, would whoever does the dmd releases consider building the release binaries with msvc? Is it Walter? Doubling the compiler's performance is significant. CTFE heavy code gets a much bigger boost too in my experience.
> 

Is an MSVC-built DMD included in the auto-tester? I think it would be prudent for that to be in place first (and using the same version of MSVC that Walter would be using).