Hi,

After a much needed rebuild of the server running various GDC-related hosted services [http://forum.dlang.org/post/zrnqcfhvyhlfjajtqebt@forum.dlang.org] - I've gotten round to updating the compiler disassembler.

http://explore.dgnu.org/

Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)

Enjoy.
Iain.

On Monday, 25 January 2016 at 23:08:32 UTC, Iain Buclaw wrote:
> Hi,
>
> After a much needed rebuild of the server running various GDC-related hosted services [http://forum.dlang.org/post/zrnqcfhvyhlfjajtqebt@forum.dlang.org] - I've gotten round to updating the compiler disassembler.
>
> http://explore.dgnu.org/
>
> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>
> Enjoy.
> Iain.

This is awesome, I think I am going to use this to finally learn some assembly. But I am not quite sure though what the output is, is it x86 or x64?

On Tue, Jan 26, 2016 at 1:08 AM, Iain Buclaw via Digitalmars-d-announce < digitalmars-d-announce@puremagic.com> wrote:

> Hi,
>
> After a much needed rebuild of the server running various GDC-related hosted services [ http://forum.dlang.org/post/zrnqcfhvyhlfjajtqebt@forum.dlang.org] - I've gotten round to updating the compiler disassembler.
>
> http://explore.dgnu.org/
>
> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>
> Enjoy.
> Iain.
>

Nice, is there a _best_ resource to understand the parameters, e.g. what is (%rip) after a symbol name?

On Tuesday, 26 January 2016 at 02:44:56 UTC, maik klein wrote:
> On Monday, 25 January 2016 at 23:08:32 UTC, Iain Buclaw wrote:
>> Hi,
>>
>> After a much needed rebuild of the server running various GDC-related hosted services [http://forum.dlang.org/post/zrnqcfhvyhlfjajtqebt@forum.dlang.org] - I've gotten round to updating the compiler disassembler.
>>
>> http://explore.dgnu.org/
>>
>> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>>
>> Enjoy.
>> Iain.
>
> This is awesome, I think I am going to use this to finally learn some assembly. But I am not quite sure though what the output is, is it x86 or x64?

The first is x64. You can switch to x86 using -m32.  However I could just add an extra compiler to do this automatically.

Iain.

On Tuesday, 26 January 2016 at 06:01:52 UTC, Rory McGuire wrote:
> On Tue, Jan 26, 2016 at 1:08 AM, Iain Buclaw via Digitalmars-d-announce < digitalmars-d-announce@puremagic.com> wrote:
>
>> Hi,
>>
>> After a much needed rebuild of the server running various GDC-related hosted services [ http://forum.dlang.org/post/zrnqcfhvyhlfjajtqebt@forum.dlang.org] - I've gotten round to updating the compiler disassembler.
>>
>> http://explore.dgnu.org/
>>
>> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>>
>> Enjoy.
>> Iain.
>>
>
> Nice, is there a _best_ resource to understand the parameters, e.g. what is (%rip) after a symbol name?

Not sure of any precise resource.  However this explains the need for %rip https://www.technovelty.org/c/position-independent-code-and-x86-64-libraries.html

On Tuesday, 26 January 2016 at 07:22:36 UTC, Iain Buclaw wrote:
> On Tuesday, 26 January 2016 at 02:44:56 UTC, maik klein wrote:
>> On Monday, 25 January 2016 at 23:08:32 UTC, Iain Buclaw wrote:
>>> Hi,
>>>
>>> After a much needed rebuild of the server running various GDC-related hosted services [http://forum.dlang.org/post/zrnqcfhvyhlfjajtqebt@forum.dlang.org] - I've gotten round to updating the compiler disassembler.
>>>
>>> http://explore.dgnu.org/
>>>
>>> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>>>
>>> Enjoy.
>>> Iain.
>>
>> This is awesome, I think I am going to use this to finally learn some assembly. But I am not quite sure though what the output is, is it x86 or x64?
>
> The first is x64. You can switch to x86 using -m32.  However I could just add an extra compiler to do this automatically.
>
> Iain.

Done.  Also made the target names more clear.

Iain.

On Monday, 25 January 2016 at 23:08:32 UTC, Iain Buclaw wrote:
> http://explore.dgnu.org/
>

Nice!
Is there a way to override the default '-Og'? It seems that now Currently I cannot see any difference

> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)

BTW, dunno how it's now, but about a year ago GDC was able to compile for AVR after removing two asserts in the frontend (checking if the pointer size is at least 32 bits or the like).

On Tuesday, 26 January 2016 at 10:17:37 UTC, Adrian Matoga wrote:

> Nice!
> Is there a way to override the default '-Og'? It seems that now Currently I cannot see any difference

Oops, pressed "Send" too quickly.
Should be: I cannot see any difference in the output when I enter various optimization options.

On Tuesday, 26 January 2016 at 10:17:37 UTC, Adrian Matoga wrote:
> On Monday, 25 January 2016 at 23:08:32 UTC, Iain Buclaw wrote:
>> http://explore.dgnu.org/
>>
>
> Nice!
> Is there a way to override the default '-Og'? It seems that now Currently I cannot see any difference
>

Ah, I left that test option in by accident.  Removed.  :-)

>> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>
> BTW, dunno how it's now, but about a year ago GDC was able to compile for AVR after removing two asserts in the frontend (checking if the pointer size is at least 32 bits or the like).

The situation should be exactly the same.  You may notice that not all targets hosted are able to compile std.stdio.writeln("Hello World").  I hope that giving awareness that these exist will help finish off the druntime ports for these platforms.

Iain

On Monday, 25 January 2016 at 23:08:32 UTC, Iain Buclaw wrote:
>
> Now supports 12 different architectures from ARM to SystemZ! (not including -m32 or any -march options)
>
I was sad that the 68k version doesn't work. :(

On an unrelated note, is is normal for write() to cause so much more ASM than writeln()?  IIRC, writeln() just wraps write(), so it seems strange that the latter would generate an order of magnitude more code than the former (looks like most of it is unicode exception stuff).

-Wyatt