I tried to build the HTML documentation with

make html

and got:

wine dmd -version=ddoc -d -c -o- ../docsrc/std.ddoc
-Df../web/2.0/phobos/std_gregorian.html std/gregorian.d
Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(25):
Error: identifier 'siginfo_t' is not defined
Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(25):
Error: siginfo_t is used as a type
Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
Error: identifier 'sigset_t' is not defined
Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
Error: sigset_t is used as a type
Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
Error: variable core.sys.posix.signal.sigaction_t.sa_mask voids have no
value
make: *** [../web/2.0/phobos/std_gregorian.html] Error 1

I reduced the problem down to this definition of std.gregorian (which is the culprit):

module std.gregorian;
import core.sys.posix.time;

So the problem is core.sys.posix.time. Since that contains "posix" in its name, does that imply it's not working on Windows? If so, you may want to insert a pragma error on version windows that leaves a nice error message behind.

Is that possible?


Andrei

This problem has been reported in d.learn.

http://lists.puremagic.com/pipermail/digitalmars-d-learn/2010-May/017087.html

Other modules also have same problem(e.g. core.sys.posix.pthread, std.windows.*).


On Sun, 23 May 2010 02:45:38 +0900, Andrei Alexandrescu <andrei at erdani.com> wrote:

> I tried to build the HTML documentation with
>
> make html
>
> and got:
>
> wine dmd -version=ddoc -d -c -o- ../docsrc/std.ddoc
> -Df../web/2.0/phobos/std_gregorian.html std/gregorian.d
> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(25):
> Error: identifier 'siginfo_t' is not defined
> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(25):
> Error: siginfo_t is used as a type
> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
> Error: identifier 'sigset_t' is not defined
> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
> Error: sigset_t is used as a type
> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
> Error: variable core.sys.posix.signal.sigaction_t.sa_mask voids have no
> value
> make: *** [../web/2.0/phobos/std_gregorian.html] Error 1
>
> I reduced the problem down to this definition of std.gregorian (which is the culprit):
>
> module std.gregorian;
> import core.sys.posix.time;
>
> So the problem is core.sys.posix.time. Since that contains "posix" in its name, does that imply it's not working on Windows? If so, you may want to insert a pragma error on version windows that leaves a nice error message behind.
>
> Is that possible?
>

Some posix modules may work on Windows if user uses INTERIX(I didn't try). So, I think module selection using version statement on user side is better than pragma insertion to each modules.

OK, I just fixed that.

Andrei

On 05/22/2010 03:00 PM, Masahiro Nakagawa wrote:
> This problem has been reported in d.learn.
>
> http://lists.puremagic.com/pipermail/digitalmars-d-learn/2010-May/017087.html
>
>
> Other modules also have same problem(e.g. core.sys.posix.pthread,
> std.windows.*).
>
>
> On Sun, 23 May 2010 02:45:38 +0900, Andrei Alexandrescu <andrei at erdani.com> wrote:
>
>> I tried to build the HTML documentation with
>>
>> make html
>>
>> and got:
>>
>> wine dmd -version=ddoc -d -c -o- ../docsrc/std.ddoc
>> -Df../web/2.0/phobos/std_gregorian.html std/gregorian.d
>> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(25):
>> Error: identifier 'siginfo_t' is not defined
>> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(25):
>> Error: siginfo_t is used as a type
>> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
>> Error: identifier 'sigset_t' is not defined
>> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
>> Error: sigset_t is used as a type
>> Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys\posix\signal.d(195):
>> Error: variable core.sys.posix.signal.sigaction_t.sa_mask voids have
>> no value
>> make: *** [../web/2.0/phobos/std_gregorian.html] Error 1
>>
>> I reduced the problem down to this definition of std.gregorian (which
>> is the culprit):
>>
>> module std.gregorian;
>> import core.sys.posix.time;
>>
>> So the problem is core.sys.posix.time. Since that contains "posix" in its name, does that imply it's not working on Windows? If so, you may want to insert a pragma error on version windows that leaves a nice error message behind.
>>
>> Is that possible?
>>
>
> Some posix modules may work on Windows if user uses INTERIX(I didn't try).
> So, I think module selection using version statement on user side is
> better than pragma insertion to each modules.
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

Windows is ostensibly posix-compliant, Ive just not bothered to add the defs.  The Windows posix subsystem ducks enough that there's really little point.

Sent from my iPhone

On May 22, 2010, at 10:45 AM, Andrei Alexandrescu <andrei at erdani.com> wrote:

> I tried to build the HTML documentation with
>
> make html
>
> and got:
>
> wine dmd -version=ddoc -d -c -o- ../docsrc/std.ddoc -Df../web/2.0/ phobos/std_gregorian.html std/gregorian.d Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys \posix\signal.d(25): Error: identifier 'siginfo_t' is not defined Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys \posix\signal.d(25): Error: siginfo_t is used as a type Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys \posix\signal.d(195): Error: identifier 'sigset_t' is not defined Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys \posix\signal.d(195): Error: sigset_t is used as a type Z:\home\andrei\dmd2\windows\bin\..\..\src\druntime\import\core\sys \posix\signal.d(195): Error: variable core.sys.posix.signal.sigaction_t.sa_mask voids have no value make: *** [../web/2.0/phobos/std_gregorian.html] Error 1
>
> I reduced the problem down to this definition of std.gregorian (which is the culprit):
>
> module std.gregorian;
> import core.sys.posix.time;
>
> So the problem is core.sys.posix.time. Since that contains "posix" in its name, does that imply it's not working on Windows? If so, you may want to insert a pragma error on version windows that leaves a nice error message behind.
>
> Is that possible?
>
>
> Andrei
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

Sean Kelly wrote:
> Windows is ostensibly posix-compliant, Ive just not bothered to add the defs.  The Windows posix subsystem ducks enough that there's really little point.
>

The Windows posix subsystem is a kludge and an extra dll, and is often treated as a poor step-daughter by Microsoft. I strongly suggest we stay away from it. Phobos should use the native Windows APIs. I also do not want to encourage its use by providing a set of posix declarations for Windows as part of Phobos.

Agreed.  I actually tried porting a large Unix app (Solaris/Linux) to Microsoft SUA once.  Simply getting it to build was a huge hassle and making it run correctly would have taken a bunch of recoding.  Between that and the abysmal command interface I really don't understand why MS bothers to maintain the thing at all.  Hm... what about Cygwin?  I suppose that might be another Windows/Posix platform?  Or is the OS "Cygwin" in that case?

On May 22, 2010, at 5:04 PM, Walter Bright wrote:
> 
> Sean Kelly wrote:
>> Windows is ostensibly posix-compliant, Ive just not bothered to add the defs.  The Windows posix subsystem ducks enough that there's really little point.
>> 
> 
> The Windows posix subsystem is a kludge and an extra dll, and is often treated as a poor step-daughter by Microsoft. I strongly suggest we stay away from it. Phobos should use the native Windows APIs. I also do not want to encourage its use by providing a set of posix declarations for Windows as part of Phobos.
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

Phobos should not in any way require or rely on Cygwin, for a long list of reasons.

Sean Kelly wrote:
> Agreed.  I actually tried porting a large Unix app (Solaris/Linux) to Microsoft SUA once.  Simply getting it to build was a huge hassle and making it run correctly would have taken a bunch of recoding.  Between that and the abysmal command interface I really don't understand why MS bothers to maintain the thing at all.  Hm... what about Cygwin?  I suppose that might be another Windows/Posix platform?  Or is the OS "Cygwin" in that case?
>
> On May 22, 2010, at 5:04 PM, Walter Bright wrote:
> 
>> Sean Kelly wrote:
>> 
>>> Windows is ostensibly posix-compliant, Ive just not bothered to add the defs.  The Windows posix subsystem ducks enough that there's really little point.
>>>
>>> 
>> The Windows posix subsystem is a kludge and an extra dll, and is often treated as a poor step-daughter by Microsoft. I strongly suggest we stay away from it. Phobos should use the native Windows APIs. I also do not want to encourage its use by providing a set of posix declarations for Windows as part of Phobos.
>> _______________________________________________
>> phobos mailing list
>> phobos at puremagic.com
>> http://lists.puremagic.com/mailman/listinfo/phobos
>> 
>
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>
>
>