Yes, to clarify, I also do not run on 64-bit linux.  My tests were 32-bit only.

-Steve



----- Original Message -----
> From:Masahiro Nakagawa <repeatedly at gmail.com>
> To:Steve Schveighoffer <schveiguy at yahoo.com>; Discuss the dmd beta releases for D <dmd-beta at puremagic.com>
> Cc:
> Sent:Wednesday, February 16, 2011 11:01 AM
> Subject:Re: [dmd-beta] Getting Beta: dmd 1.067 and 2.052 beta
> 
> MessagePack, new socket and Mustache template engine too. But, I don't run on 64bit Linux.
> 
> 
> Masahiro
> 
> On Wed, 16 Feb 2011 23:28:03 +0900, Steve Schveighoffer <schveiguy at yahoo.com> wrote:
> 
> > Compiles and passes unit tests with dcollections.
> > 

So am I really the only one that's tested DMD64 on any real-world code besides Phobos and druntime so far?

On Wed, Feb 16, 2011 at 1:11 PM, Steve Schveighoffer <schveiguy at yahoo.com>wrote:

> Yes, to clarify, I also do not run on 64-bit linux.  My tests were 32-bit only.
>
> -Steve
>
>
>
> ----- Original Message -----
> > From:Masahiro Nakagawa <repeatedly at gmail.com>
> > To:Steve Schveighoffer <schveiguy at yahoo.com>; Discuss the dmd beta
> releases for D <dmd-beta at puremagic.com>
> > Cc:
> > Sent:Wednesday, February 16, 2011 11:01 AM
> > Subject:Re: [dmd-beta] Getting Beta: dmd 1.067 and 2.052 beta
> >
> > MessagePack, new socket and Mustache template engine too. But, I don't run on 64bit Linux.
> >
> >
> > Masahiro
> >
> > On Wed, 16 Feb 2011 23:28:03 +0900, Steve Schveighoffer <schveiguy at yahoo.com> wrote:
> >
> > > Compiles and passes unit tests with dcollections.
> > >
>
>
>
> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/phobos/attachments/20110216/6f877ca1/attachment.html>

Depends on your definition of real world.  A whole lot of the dmd test suite is distilled past problems from real world code.  What is usually called regression tests.  The 3 test suites aren't small.  Through those, a huge number of issues similar to yours have been found and fixed.  There's a little app in the dmd test suite that actually drives the testing.  It's small, but it's real worldish.

That said, I don't have any production d code of any flavor.

On 2/16/2011 10:14 AM, David Simcha wrote:
> So am I really the only one that's tested DMD64 on any real-world code besides Phobos and druntime so far?
> 
> On Wed, Feb 16, 2011 at 1:11 PM, Steve Schveighoffer <schveiguy at yahoo.com <mailto:schveiguy at yahoo.com>> wrote:
> 
>     Yes, to clarify, I also do not run on 64-bit linux.  My tests were 32-bit only.
> 
>     -Steve
> 
> 
> 
>     ----- Original Message -----
>     > From:Masahiro Nakagawa <repeatedly at gmail.com <mailto:repeatedly at gmail.com>>
>     > To:Steve Schveighoffer <schveiguy at yahoo.com <mailto:schveiguy at yahoo.com>>; Discuss the dmd beta releases for D
>     <dmd-beta at puremagic.com <mailto:dmd-beta at puremagic.com>>
>     > Cc:
>     > Sent:Wednesday, February 16, 2011 11:01 AM
>     > Subject:Re: [dmd-beta] Getting Beta: dmd 1.067 and 2.052 beta
>     >
>     > MessagePack, new socket and Mustache template engine too.
>     > But, I don't run on 64bit Linux.
>     >
>     >
>     > Masahiro
>     >
>     > On Wed, 16 Feb 2011 23:28:03 +0900, Steve Schveighoffer
>     > <schveiguy at yahoo.com <mailto:schveiguy at yahoo.com>> wrote:
>     >
>     > > Compiles and passes unit tests with dcollections.
>     > >
> 

I expect it will take at least a couple releases before the 64 bit compiler is reasonably solid. Few will actually try it until it is released. That's just the way of things.

One difficulty is the D test suite does not have a lot in it to test the optimizer/code generator. That's because the test suite consists of tests for things that failed in the past, and the optimizer/code generator was already solid from the C++ usage. The C++ compiler test suite does test the optimizer/code generator, but it isn't 64 bit.

Brad Roberts wrote:
> Depends on your definition of real world.  A whole lot of the dmd test suite is distilled past problems from real world code.  What is usually called regression tests.  The 3 test suites aren't small.  Through those, a huge number of issues similar to yours have been found and fixed.  There's a little app in the dmd test suite that actually drives the testing.  It's small, but it's real worldish.
>
> That said, I don't have any production d code of any flavor.
>
> On 2/16/2011 10:14 AM, David Simcha wrote:
> 
>> So am I really the only one that's tested DMD64 on any real-world code besides Phobos and druntime so far?
>>
>> 

Really?  Despite the one 64 issue that I just filed recently that hasn't been fixed, I actually find it pretty usable and used it for a little real work today.  Maybe that's because I went through all the library code and a bunch of the application code I use with a fine-toothed comb and filed every bug report I could find and all but one have been fixed.  I'm not sure how well the solidness on this code generalizes to code where noone has done this yet.

On Wed, Feb 16, 2011 at 2:00 PM, Walter Bright <walter at digitalmars.com>wrote:

> I expect it will take at least a couple releases before the 64 bit compiler is reasonably solid. Few will actually try it until it is released. That's just the way of things.
>
> One difficulty is the D test suite does not have a lot in it to test the optimizer/code generator. That's because the test suite consists of tests for things that failed in the past, and the optimizer/code generator was already solid from the C++ usage. The C++ compiler test suite does test the optimizer/code generator, but it isn't 64 bit.
>
>
> Brad Roberts wrote:
>
>> Depends on your definition of real world.  A whole lot of the dmd test
>> suite is distilled past problems from real world
>> code.  What is usually called regression tests.  The 3 test suites aren't
>> small.  Through those, a huge number of issues
>> similar to yours have been found and fixed.  There's a little app in the
>> dmd test suite that actually drives the
>> testing.  It's small, but it's real worldish.
>>
>> That said, I don't have any production d code of any flavor.
>>
>> On 2/16/2011 10:14 AM, David Simcha wrote:
>>
>>
>>> So am I really the only one that's tested DMD64 on any real-world code besides Phobos and druntime so far?
>>>
>>>
>>>
>> _______________________________________________
> phobos mailing list
> phobos at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/phobos
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.puremagic.com/pipermail/phobos/attachments/20110216/5164e4e3/attachment.html>

It's a major new area of functionality, it's gonna have lots of interesting bugs.  With luck the reports will be as good as yours.  With more luck they'll be as easy to fix as these have been.  I'm much more confident in the latter than the former. :)

For what it's worth, there's still a couple broken phobos tests for 64 bit that need to be isolated and fixed.  If you're feeling generous, look in posix.mak for DISABLED and go nuts.

Later,
Brad

On Wed, 16 Feb 2011, David Simcha wrote:

> Really?  Despite the one 64 issue that I just filed recently that hasn't been fixed, I actually find it pretty usable and used it for a little real work today.  Maybe that's because I went through all the library code and a bunch of the application code I use with a fine-toothed comb and filed every bug report I could find and all but one have been fixed.  I'm not sure how well the solidness on this code generalizes to code where noone has done this yet.
> 
> On Wed, Feb 16, 2011 at 2:00 PM, Walter Bright <walter at digitalmars.com>wrote:
> 
> > I expect it will take at least a couple releases before the 64 bit compiler is reasonably solid. Few will actually try it until it is released. That's just the way of things.
> >
> > One difficulty is the D test suite does not have a lot in it to test the optimizer/code generator. That's because the test suite consists of tests for things that failed in the past, and the optimizer/code generator was already solid from the C++ usage. The C++ compiler test suite does test the optimizer/code generator, but it isn't 64 bit.
> >
> >
> > Brad Roberts wrote:
> >
> >> Depends on your definition of real world.  A whole lot of the dmd test
> >> suite is distilled past problems from real world
> >> code.  What is usually called regression tests.  The 3 test suites aren't
> >> small.  Through those, a huge number of issues
> >> similar to yours have been found and fixed.  There's a little app in the
> >> dmd test suite that actually drives the
> >> testing.  It's small, but it's real worldish.
> >>
> >> That said, I don't have any production d code of any flavor.
> >>
> >> On 2/16/2011 10:14 AM, David Simcha wrote:
> >>
> >>
> >>> So am I really the only one that's tested DMD64 on any real-world code besides Phobos and druntime so far?
> >>>
> >>>
-------------- next part --------------
_______________________________________________
phobos mailing list
phobos at puremagic.com
http://lists.puremagic.com/mailman/listinfo/phobos