Update to fix:

http://www.digitalmars.com/d/archives/digitalmars/D/announce/Do_not_use_2.041_17841.html http://d.puremagic.com/issues/show_bug.cgi?id=3930

You can download the software as a zip file from:


http://download.digitalmars.com/dmd2beta.zip
userid: customer
passwd: download7

Looks good.  Only thing is that the changelog doesn't contain any updates, and the docs don't reflect the new names of the array functions I added.  It appears maybe that the docs were not regenerated.

-Steve




----- Original Message ----
> From: Walter Bright <walter at digitalmars.com>
> To: Discuss the dmd beta releases for D <dmd-beta at puremagic.com>
> Sent: Sun, March 14, 2010 5:12:50 PM
> Subject: [dmd-beta] beta for dmd 2.042
> 
> Update to fix:

http://www.digitalmars.com/d/archives/digitalmars/D/announce/Do_not_use_2.041_17841.html http://d.puremagic.com/issues/show_bug.cgi?id=3930

You
> can download the software as a zip file from:


http://download.digitalmars.com/dmd2beta.zip
userid:
> customer
passwd:
> download7
_______________________________________________
dmd-beta mailing
> list

> href="mailto:dmd-beta at puremagic.com">dmd-beta at puremagic.com
http://lists.puremagic.com/mailman/listinfo/dmd-beta

This is an awesome release.
One thing though -- the Phobos unittests don't pass. The unittest
around socket.d 292 fails on Windows.
Admittedly, the unittests failed on the previous release, too, but I
don't think we should do it again. At the very least we should do
something like:

unittest
{
	Protocol proto = new Protocol;
+    version(Windows)    {
+        pragma(msg, " --- socket.d(" ~ __LINE__ ~ ") broken test ---");
+	}	else	{
	assert(proto.getProtocolByType(ProtocolType.TCP));


On 14 March 2010 22:12, Walter Bright <walter at digitalmars.com> wrote:
> Update to fix:
>
> http://www.digitalmars.com/d/archives/digitalmars/D/announce/Do_not_use_2.041_17841.html http://d.puremagic.com/issues/show_bug.cgi?id=3930
>
> You can download the software as a zip file from:
>
>
> http://download.digitalmars.com/dmd2beta.zip
> userid: customer
> passwd: download7
> _______________________________________________
> dmd-beta mailing list
> dmd-beta at puremagic.com
> http://lists.puremagic.com/mailman/listinfo/dmd-beta
>

Yeah, I ran into that, too. I don't know what it does or how it's supposed to work, so I checked in the workaround you suggest.

Don Clugston wrote:
> This is an awesome release.
> One thing though -- the Phobos unittests don't pass. The unittest
> around socket.d 292 fails on Windows.
> Admittedly, the unittests failed on the previous release, too, but I
> don't think we should do it again. At the very least we should do
> something like:
>
> unittest
> {
> 	Protocol proto = new Protocol;
> +    version(Windows)    {
> +        pragma(msg, " --- socket.d(" ~ __LINE__ ~ ") broken test ---");
> +	}	else	{
> 	assert(proto.getProtocolByType(ProtocolType.TCP));
>
> 

Doesn't concern the release directly, but I ran into it while testing and I don't think it's worth its own bugzilla issue.

http://www.digitalmars.com/d/2.0/operatoroverloading.html#OpAssign

"a op= b

is rewritten as:
a.opOpAssign!("op")(b)"

is misleading. I deduced that op doesn't include "=" but it does.

> One thing though -- the Phobos unittests don't pass. The unittest
> around socket.d 292 fails on Windows.
> Admittedly, the unittests failed on the previous release, too, but I
> don't think we should do it again. At the very least we should do
> something like:
>

For me, also these fail:

D:\dmd\windows\bin\..\..\src\phobos\std\random.d(1186): Error: function
std.random.rand is deprecated
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(169): Error: cannot
evaluate primeFactorsOnly(100LU) at compile time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(169): Error: static
assert  (primeFactorsOnly(100LU) == 10LU) is not evaluatable at compile
time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(171): Error: cannot
evaluate primeFactorsOnly(11LU) at compile time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(171): Error: static
assert  (primeFactorsOnly(11LU) == 11LU) is not evaluatable at compile time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(172): Error: cannot
evaluate primeFactorsOnly(622545LU) at compile time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(172): Error: static
assert  (primeFactorsOnly(622545LU) == 1155LU) is not evaluatable at
compile time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(173): Error: cannot
evaluate primeFactorsOnly(258LU) at compile time
D:\dmd\windows\bin\..\..\src\phobos\std\random.d(173): Error: static
assert  (primeFactorsOnly(258LU) == 258LU) is not evaluatable at compile
time

  --- std.regex(3318) broken test ---
D:\dmd\windows\bin\..\..\src\phobos\std\string.d(4058): Error: function
std.string.find is deprecated

Third issue:
Somebody messed up std.perf. Doesn't measure anything anymore.

This is the new version, line 95:

interval_t freq;
if (QueryPerformanceFrequency(&freq))
	sm_freq =   freq;
else
	sm_freq =   1000;
sm_fn	=   &_qtc;


sm_fn isn't assigned and the brackets aren't correct (right?)

This is the old version:

if(QueryPerformanceFrequency(&sm_freq))
{
	sm_fn	=   &_qpc;
}
else
{
	sm_freq =   1000;
	sm_fn	=   &_qtc;
}


I don't even know why another variable is used. If QueryPerformanceFrequency succeeds, sm_freq is already set; else it gets overwritten in the else clause.

> I don't even know why another variable is used. If QueryPerformanceFrequency succeeds, sm_freq is already set; else it gets overwritten in the else clause.
>

Ah never mind. Cause it's now const and not an lvalue.
The following is the corrected code. Tested it, works fine again:


interval_t freq;
if (QueryPerformanceFrequency(&freq))
{
	sm_freq =   freq;
	sm_fn	=   &_qpc;
}
else
{
	sm_freq =   1000;
	sm_fn	=   &_qtc;
}

Trass3r wrote:
> Doesn't concern the release directly, but I ran into it while testing and I don't think it's worth its own bugzilla issue.
>
> http://www.digitalmars.com/d/2.0/operatoroverloading.html#OpAssign
>
> "a op= b
>
> is rewritten as:
> a.opOpAssign!("op")(b)"
>
> is misleading. I deduced that op doesn't include "=" but it does.
>

You're right. Will fix.

Trass3r wrote:
> Third issue:
> Somebody messed up std.perf. Doesn't measure anything anymore.
>

That would be me. I failed to spot the difference between _qpc and _qtc and thought I'd "optimize" the code. Thanks for setting the code straight.