FWIW, I get the same error Walter does, when trying to build the documentation.  'make -f posix.mak html' runs the Windows DMD through wine.

-Lars



On Tue, 2010-09-14 at 00:36 -0700, Brad Roberts wrote:
> On 9/14/2010 12:21 AM, Walter Bright wrote:
> > 
> > Thanks! I updated, but the compile failure persists:
> > 
> > H:\mars\phobos>..\dmd -c std\file.d
> > std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue
> > 
> 
> Ok.. it's something that's different for your env from mine.. my win32 build passes just fine.  It's completed a build cycle since dsource came back up and it built and passes all the tests just fine.
> 
> The script I use to do the phobos build:
>    http://github.com/braddr/d-tester/blob/master/client/src/do_build_phobos.sh
> 
> The log for the most recent win32 phobos build:
>    http://d.puremagic.com/test-results/test_data.ghtml?dataid=2346
> 
> Every build cycle is done from a completely fresh checkout straight from
> dsource.  The only changes that are applied on top of it come from:
>    http://github.com/braddr/d-tester/blob/master/client/src/do_fixup.sh
>    http://github.com/braddr/d-tester/blob/master/client/src/patch-dmd-win32.mak
>    http://github.com/braddr/d-tester/blob/master/client/src/sc.ini
> 
> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

On 09/14/2010 02:21 AM, Walter Bright wrote:
> Thanks! I updated, but the compile failure persists:
>
> H:\mars\phobos>..\dmd -c std\file.d
> std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue

I assumed it must be the new compiler's changes in typechecking, but I just got the latest compiler, copied it over the older one, then:

make -f posix.mak clean
make -f posix.mak unittest

and it all worked.

FWIW, line 3200 is std/algorithm is:

endsWith(alias pred = "a == b", Range, Ranges...)

I assume it must be a Windows-specific error, but I can't check because I don't have a recent dmd.exe. Walter, could you please email me one? Don't forget to rename the extension.

One conclusion for the future is that we need to be able to build dmd on wine. That way we all get to build on wine using the latest compiler.


Andrei

The problem is in the Windows-only code in  std.path, dirname(), line 426.
            if (uint withWhat = s.endsWith(sep, altsep, drvsep))
Don't know why it fails, though.



On 14 September 2010 16:57, Andrei Alexandrescu <andrei at erdani.com> wrote:
> On 09/14/2010 02:21 AM, Walter Bright wrote:
>>
>> Thanks! I updated, but the compile failure persists:
>>
>> H:\mars\phobos>..\dmd -c std\file.d
>> std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue

Must be again the fixed-length string. I need to look into this, but I'm also caught up at work. If anyone else wants to look at this before tonight, please announce.

Andrei

On 9/14/10 10:53 CDT, Don Clugston wrote:
> The problem is in the Windows-only code in  std.path, dirname(), line 426.
>              if (uint withWhat = s.endsWith(sep, altsep, drvsep))
> Don't know why it fails, though.
>
>
>
> On 14 September 2010 16:57, Andrei Alexandrescu<andrei at erdani.com>  wrote:
>> On 09/14/2010 02:21 AM, Walter Bright wrote:
>>>
>>> Thanks! I updated, but the compile failure persists:
>>>
>>> H:\mars\phobos>..\dmd -c std\file.d
>>> std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos

Don Clugston <dclugston at googlemail.com> wrote:
> The problem is in the Windows-only code in  std.path, dirname(), line 426.
>             if (uint withWhat = s.endsWith(sep, altsep, drvsep))
> Don't know why it fails, though.

Filed a bug:
- Static-to-dynamic converted manifest constant array gets
  non-converted type in static/constraint if
http://d.puremagic.com/issues/show_bug.cgi?id=4866

The type of altsep was wrong in the constraint of endsWith().


Shin

On 14 September 2010 20:18, Shin Fujishiro <rsinfu at gmail.com> wrote:
> Don Clugston <dclugston at googlemail.com> wrote:
>> The problem is in the Windows-only code in ?std.path, dirname(), line 426.
>> ? ? ? ? ? ? if (uint withWhat = s.endsWith(sep, altsep, drvsep))
>> Don't know why it fails, though.
>
> Filed a bug:
> - Static-to-dynamic converted manifest constant array gets
> ?non-converted type in static/constraint if
> http://d.puremagic.com/issues/show_bug.cgi?id=4866
>
> The type of altsep was wrong in the constraint of endsWith().
>
>
> Shin

Thanks Shin, you're a legend!

Hrm, so it's doc generation that's causing it?  Is that part of the default target on win32 but not posix?  That wouldnt explain why I'm not seeing the problem on win32 though.

What am I doing or not doing that's making me not see the problem?  I want to increase my test coverage to catch this issue.

Confused,
Brad

On Tue, 14 Sep 2010, Lars Tandle Kyllingstad wrote:

> Date: Tue, 14 Sep 2010 11:40:27 +0200
> From: Lars Tandle Kyllingstad <lars at kyllingen.net>
> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
> To: Discuss the phobos library for D <phobos at puremagic.com>
> Subject: Re: [phobos] build failure in std.file
> 
> FWIW, I get the same error Walter does, when trying to build the documentation.  'make -f posix.mak html' runs the Windows DMD through wine.
> 
> -Lars
> 
> 
> 
> On Tue, 2010-09-14 at 00:36 -0700, Brad Roberts wrote:
> > On 9/14/2010 12:21 AM, Walter Bright wrote:
> > > 
> > > Thanks! I updated, but the compile failure persists:
> > > 
> > > H:\mars\phobos>..\dmd -c std\file.d
> > > std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue
> > > 
> > 
> > Ok.. it's something that's different for your env from mine.. my win32 build passes just fine.  It's completed a build cycle since dsource came back up and it built and passes all the tests just fine.
> > 
> > The script I use to do the phobos build:
> >    http://github.com/braddr/d-tester/blob/master/client/src/do_build_phobos.sh
> > 
> > The log for the most recent win32 phobos build:
> >    http://d.puremagic.com/test-results/test_data.ghtml?dataid=2346
> > 
> > Every build cycle is done from a completely fresh checkout straight from
> > dsource.  The only changes that are applied on top of it come from:
> >    http://github.com/braddr/d-tester/blob/master/client/src/do_fixup.sh
> >    http://github.com/braddr/d-tester/blob/master/client/src/patch-dmd-win32.mak
> >    http://github.com/braddr/d-tester/blob/master/client/src/sc.ini
> > 
> > 
> > _______________________________________________
> > 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
> 

The latest checkin to std.path fixes the problem.

Andrei Alexandrescu wrote:
> On 09/2010 02:21 AM, Walter Bright wrote:
>> Thanks! I updated, but the compile failure persists:
>>
>> H:\mars\phobos>..\dmd -c std\file.d
>> std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue
>
> I assumed it must be the new compiler's changes in typechecking, but I just got the latest compiler, copied it over the older one, then:
>
> make -f posix.mak clean
> make -f posix.mak unittest
>
> and it all worked.
>

Since it happened with:

  dmd -c std/file

it wasn't the doc generation.

Brad Roberts wrote:
> Hrm, so it's doc generation that's causing it?  Is that part of the default target on win32 but not posix?  That wouldnt explain why I'm not seeing the problem on win32 though.
>
> What am I doing or not doing that's making me not see the problem?  I want to increase my test coverage to catch this issue.
>
> Confused,
> Brad
>
> On Tue, 14 Sep 2010, Lars Tandle Kyllingstad wrote:
>
> 
>> Date: Tue, 14 Sep 2010 11:40:27 +0200
>> From: Lars Tandle Kyllingstad <lars at kyllingen.net>
>> Reply-To: Discuss the phobos library for D <phobos at puremagic.com>
>> To: Discuss the phobos library for D <phobos at puremagic.com>
>> Subject: Re: [phobos] build failure in std.file
>>
>> FWIW, I get the same error Walter does, when trying to build the documentation.  'make -f posix.mak html' runs the Windows DMD through wine.
>>
>> -Lars
>>
>>
>>
>> On Tue, 2010-09-14 at 00:36 -0700, Brad Roberts wrote:
>> 
>>> On 9/14/2010 12:21 AM, Walter Bright wrote:
>>> 
>>>> Thanks! I updated, but the compile failure persists:
>>>>
>>>> H:\mars\phobos>..\dmd -c std\file.d
>>>> std\algorithm.d(3200): Error: cast(string)_param_1 is not an lvalue
>>>>
>>>> 
>>> Ok.. it's something that's different for your env from mine.. my win32 build passes just fine.  It's completed a build cycle since dsource came back up and it built and passes all the tests just fine.
>>>
>>> The script I use to do the phobos build:
>>>    http://github.com/braddr/d-tester/blob/master/client/src/do_build_phobos.sh
>>>
>>> The log for the most recent win32 phobos build:
>>>    http://d.puremagic.com/test-results/test_data.ghtml?dataid=2346
>>>
>>> Every build cycle is done from a completely fresh checkout straight from
>>> dsource.  The only changes that are applied on top of it come from:
>>>    http://github.com/braddr/d-tester/blob/master/client/src/do_fixup.sh
>>>    http://github.com/braddr/d-tester/blob/master/client/src/patch-dmd-win32.mak
>>>    http://github.com/braddr/d-tester/blob/master/client/src/sc.ini
>>>
>>>
>>> _______________________________________________
>>> 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
>>
>> 
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>
>
> 

On 14 September 2010 21:07, Brad Roberts <braddr at puremagic.com> wrote:
> Hrm, so it's doc generation that's causing it? ?Is that part of the default target on win32 but not posix? ?That wouldnt explain why I'm not seeing the problem on win32 though.
>
> What am I doing or not doing that's making me not see the problem? ?I want to increase my test coverage to catch this issue.

For me, make -fwin32.mak unittest did not hit the problem. It only showed up with dmd -c std\file.