On Mon, 14 Oct 2013 23:13:25 +0200
"monarch_dodra" <monarchdodra@gmail.com> wrote:

> On Monday, 14 October 2013 at 13:25:23 UTC, Benjamin Thaut wrote:
> > I'm also getting random missing symbol linker errors with both
> > dmd 2.063.2 and dmd 2.064. But only on 32-bit windows. On
> > 64-bit windows it works fine.
> > This is really frustrating...
> 
> I've encountered this too. I'll try to reduce, but the test case isn't easy.

I've been bit by a similar (same?) issue.

What I didn't realize is that DMD *doesn't* pass the LIB directories (from sc.ini) into optlink. Optlink *itself* reads sc.ini. So if the optlink being run isn't in the same directory as dmd.exe, then optlink may end up grabbing the wrong sc.ini and therefore the wrong Phobos as well. Hence, weird linker errors for Win32.

Relevant "issue": http://d.puremagic.com/issues/show_bug.cgi?id=10729

On 2013-10-13 00:16, Walter Bright wrote:
> http://ftp.digitalmars.com/dmd2beta.zip
>
> Current list of regressions:

Another one: http://d.puremagic.com/issues/show_bug.cgi?id=11268

-- 
/Jacob Carlborg

Am 14.10.2013 23:19, schrieb Walter Bright:
> On 10/14/2013 6:25 AM, Benjamin Thaut wrote:
>> I'm also getting random missing symbol linker errors with both dmd
>> 2.063.2 and
>> dmd 2.064. But only on 32-bit windows. On 64-bit windows it works fine.
>> This is really frustrating...
>
> Is it possible you are linking together code compiled with different
> command line -version or -debug switches?

I dind't change anything on the build setup. And it worked with dmd 2.062. Is there now different mangeling depending on the -version and -debug statements?

On 10/15/2013 1:50 AM, Benjamin Thaut wrote:
> Am 14.10.2013 23:19, schrieb Walter Bright:
>> On 10/14/2013 6:25 AM, Benjamin Thaut wrote:
>>> I'm also getting random missing symbol linker errors with both dmd
>>> 2.063.2 and
>>> dmd 2.064. But only on 32-bit windows. On 64-bit windows it works fine.
>>> This is really frustrating...
>>
>> Is it possible you are linking together code compiled with different
>> command line -version or -debug switches?
>
> I dind't change anything on the build setup. And it worked with dmd 2.062. Is
> there now different mangeling depending on the -version and -debug statements?

dmd now assumes that templates instantiated by a library module are actually in the library.

But if code is turned on and off with -version or -debug command line switches, and different switches are used to compile the library than the importer, then the templates instantiations may not be in the library.

Am 15.10.2013 11:25, schrieb Walter Bright:
> On 10/15/2013 1:50 AM, Benjamin Thaut wrote:
>> Am 14.10.2013 23:19, schrieb Walter Bright:
>>> On 10/14/2013 6:25 AM, Benjamin Thaut wrote:
>>>> I'm also getting random missing symbol linker errors with both dmd
>>>> 2.063.2 and
>>>> dmd 2.064. But only on 32-bit windows. On 64-bit windows it works fine.
>>>> This is really frustrating...
>>>
>>> Is it possible you are linking together code compiled with different
>>> command line -version or -debug switches?
>>
>> I dind't change anything on the build setup. And it worked with dmd
>> 2.062. Is
>> there now different mangeling depending on the -version and -debug
>> statements?
>
> dmd now assumes that templates instantiated by a library module are
> actually in the library.
>
> But if code is turned on and off with -version or -debug command line
> switches, and different switches are used to compile the library than
> the importer, then the templates instantiations may not be in the library.

The funny thing is, its not a template. Nothing fancy at all. Just a struct with two members. And the linker complains that the __init member of that struct is missing.

Error 42: Symbol Undefined _D6thBase6plugin8ScanPair6__initZ

Also the library and importer are compiled with exactly the same -debug and -version switches.

I did setup a dustmite reduce process but its going to take a few hours for that to complete.

Kind Regards
Benjamin Thaut

On 10/13/13, Tourist <gravatar@gravatar.com> wrote:
> I'm wondering whether there will be the nifty changelog like it
> was for 2.063?
> Andrej? :D

We'll see if someone else volunteers to do it. I'm not doing it out of protest.

On 10/15/13 7:15 PM, Andrej Mitrovic wrote:
> On 10/13/13, Tourist <gravatar@gravatar.com> wrote:
>> I'm wondering whether there will be the nifty changelog like it
>> was for 2.063?
>> Andrej? :D
>
> We'll see if someone else volunteers to do it. I'm not doing it out of protest.

What are you protesting against?

Andrei

On 10/16/13, Andrei Alexandrescu <SeeWebsiteForEmail@erdani.org> wrote:
> What are you protesting against?

Walter.

On Saturday, 12 October 2013 at 22:16:13 UTC, Walter Bright wrote:
> http://ftp.digitalmars.com/dmd2beta.zip
>
> Current list of regressions:
>
> http://d.puremagic.com/issues/buglist.cgi?query_format=advanced&bug_severity=regression&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED
>
> This isn't a release candidate, in particular the documentation needs work, but we need to shake the tree for any undetected regressions.
>
> Further beta announcements go in the dmd-beta mailing list.
>
> Note that this release contains:
>
> 29 enhancements
> 307 dmd bugs fixed
> 14 druntime bugs fixed
> 73 phobos bugs fixed

Hum I have several regression is SDC's test suite. I have to investigate more to fix the code or submit bug report. It looks related to AA. What are the changes that affect AA in this new release ?

On Wednesday, 16 October 2013 at 13:34:04 UTC, Andrej Mitrovic wrote:
> On 10/16/13, Andrei Alexandrescu <SeeWebsiteForEmail@erdani.org> wrote:
>> What are you protesting against?
>
> Walter.

The last change log was awesome.
I vote to get rid of Walter. :)