[phobos] std.complex not compiled into phobos.lib on Windows

[Don Clugston]

On 11 May 2010 10:13, Lars Tandle Kyllingstad <lars at kyllingen.net> wrote:
> On Tue, 2010-05-11 at 10:01 +0200, Don Clugston wrote:
>> On 11 May 2010 09:11, Lars Tandle Kyllingstad <lars at kyllingen.net> wrote:
>> > Based on the following D.learn thread it would appear that std.complex
>> > isn't compiled into phobos.lib on Windows:
>> >
>> > http://www.digitalmars.com/webnews/newsgroups.php?art_group=digitalmars.D.learn&article_id=19695
>> >
>> > I have noted that complex.d is missing from the SRCS variable defined at
>> > lines 97-137 of win32.mak, and I assume that's the cause of the problem.
>> >
>> > However, I am completely unable to get Phobos to compile on Windows so I
>> > can't know for certain whether adding it fixes the problem.  (Phobos
>> > won't build without druntime, and apparently I can't build druntime
>> > without some 'masm386' program which I don't have.)
>>
>> I'm sick of that recurring. Neither file has changed for several
>> years, but at some point minit.asm was moved, and it got a different
>> date on it. I think it's fixed in the download but not in svn. Or vice
>> versa. You can also fix it by deleting minit.asm and copying it from
>> D1 or from an old D2.
>
> Perhaps it would be better to just remove mention of minit.asm from the
> makefile once and for all?  masm386 doesn't seem to be readily
> available, so most users won't be able to build it anyway.

masm386 is an easily available download. The problem with minit.asm is
just a one-off that's taken a long time to be resolved. I think it's
fixed now, and you've only encountered it by having parts of an old
repository still around.
The problem also resurfaced once when an incorrect refactoring of
win32.mak was made.

>> > Can someone confirm that adding complex.d to SRCS fixes the issue?
>>
>> Yes, that's correct. I'll fix it as soon as dsource comes back on line.

Actually I'll leave it to Walter.