Walter Bright wrote:
> Mostly bug fixes for CTFE. Added library switches at Tango's request.
> 
> http://www.digitalmars.com/d/1.0/changelog.html http://ftp.digitalmars.com/dmd.1.021.zip
> 
> http://www.digitalmars.com/d/changelog.html http://ftp.digitalmars.com/dmd.2.004.zip

*ahem*

 H A L L E L U J A !

Oh you've made me a very happy boy. :)  The multiple module ctors/dtors thing is *very* welcome.  I'll have to poke around the new 2.0 stuff, too.

The only thing left that would allow me to ditch my current, let's call it, "insane" compiler set up would be a switch to specify a different sc.ini file.

But none the less, thanks very much for these! :)

	-- Daniel

Walter Bright wrote:
> BLS wrote:
>> However. I have no idea which advantages this feature really has.
> 
> In a long module, you can organize the static constructor code in a way that makes sense, rather than being forced to put it all in one place. It also makes it practical to mixin code that requires static construction.

Badass, it is good to see this rough edge get smoothed.  Thank you Walter!

I think I used to have some hack where I would wrap each static ctor in its own class, and somehow this would make it work.  I'm not sure if that's correct or not though.

Walter Bright wrote:
> Mostly bug fixes for CTFE. Added library switches at Tango's request.
> 
> http://www.digitalmars.com/d/1.0/changelog.html
> http://ftp.digitalmars.com/dmd.1.021.zip
> 
> http://www.digitalmars.com/d/changelog.html
> http://ftp.digitalmars.com/dmd.2.004.zip

Sweet, I like it.  Thank you!!111

One thing though, when I run this:
dmd -defaultlib
it outputs this:
Error: unrecognized switch '-defaultlib'

Same with -debuglib.
Am I missing something?

Also, the -defaultlib and -debuglib switches does not appear in the dmd usage display on the commmandline.

negerns

Chad J wrote:
> One thing though, when I run this:
> dmd -defaultlib
> it outputs this:
> Error: unrecognized switch '-defaultlib'

Try:
	dmd -defaultlib=foo test.d

Chad J wrote:
> Walter Bright wrote:
>> Mostly bug fixes for CTFE. Added library switches at Tango's request.
>>
>> http://www.digitalmars.com/d/1.0/changelog.html
>> http://ftp.digitalmars.com/dmd.1.021.zip
>>
>> http://www.digitalmars.com/d/changelog.html
>> http://ftp.digitalmars.com/dmd.2.004.zip
> 
> Sweet, I like it.  Thank you!!111
> 
> One thing though, when I run this:
> dmd -defaultlib
> it outputs this:
> Error: unrecognized switch '-defaultlib'
> 
> Same with -debuglib.
> Am I missing something?

Try running dmd by itself and checking the version.  I'll bet you downloaded dmd.zip which points to 1.016 still, not 1.021.

Walter Bright wrote:
> Chad J wrote:
>> One thing though, when I run this:
>> dmd -defaultlib
>> it outputs this:
>> Error: unrecognized switch '-defaultlib'
> 
> Try:
>     dmd -defaultlib=foo test.d

Ah, that works.  As negerns mentioned, this doesn't show in the dmd usage info.  If it did, that would probably help ;)

On Wed, 05 Sep 2007 12:05:07 -0700, Walter Bright <newshound1@digitalmars.com> wrote:

>Mostly bug fixes for CTFE. Added library switches at Tango's request.
>
>http://www.digitalmars.com/d/1.0/changelog.html http://ftp.digitalmars.com/dmd.1.021.zip
>
>http://www.digitalmars.com/d/changelog.html http://ftp.digitalmars.com/dmd.2.004.zip

Thanks a lot!

yidabu wrote:
> build every program, cause:
> Compile error: QuadPart is not a member  of LARGE_INTEGER 

I've just encountered the same error ("Error: 'QuadPart' is not a member of 'LARGE_INTEGER'") when I tried to compile the WindowsAPI sources. Could be related to: http://d.puremagic.com/issues/show_bug.cgi?id=1473

DMD 1.021, Windows Vista

LLAP,
Sascha

Walter Bright wrote:
> Brad Roberts wrote:
>> The "Download latest D 2.0 alpha D compiler for Win32 and x86 linux" link on http://www.digitalmars.com/d/changelog.html still points to 2.002.
>>
>> Similarly, though at least labeled, the 1.0 changelog still points to 1.016, now 5 versions behind?  Now that the 1.0 code line is no longer receiving anything other than bug fixes, is there really the need to distinguish between the latest 1.0 release and some other really stable 1.0 release?
> 
> I think their is still a need, as there's always a risk I break something with a new release, even if it's just bug fixes.

1.020 seemed to be stable. Like 1.016, it was around for a long time, and therefore particularly well tested. There were some great bug fixes in 1.018 and 1.019.

There's that substantive change about .init which happened in 1.017. If that's permanent, it'd be good to stop further development relying on the old behaviour.
I think we need a policy for when the 'stable version' should be updated.


Also, I don't see any mention of delimited string literals in the changelog. <g>