compile on openbsd: mostly ifdefs and missing includes,...

compile on openbsd: mostly ifdefs and missing includes,
a few small changes but all ifdef'ed to openbsd

[pvalchev, 2009-06-09 23:20:27]

These are most of the changes for compiling src/base_unittest on OpenBSD. As FYI
I'll have to clean up some of the other stuff - in the meantime ran into a
problem with our toolchain's wchar support being broken so working on a fix for
that before this effort can resume... (linking the unittest failed revealing
some issues)

[tony, 2009-06-10 18:42:04]

Evan is on vacation this week so I'm taking this.

http://codereview.chromium.org/118469/diff/1032/1045
File base/clipboard.h (right):

http://codereview.chromium.org/118469/diff/1032/1045#newcode24
Line 24: #if defined(OS_LINUX) || defined(OS_OPENBSD)
After some discussion with some others working on linux, we've decided that we
should add a BASE_GTK define to build_config.h.  You could imagine some day
there being a BASE_QT that would be the other side of this.  A lot of the
OS_LINUX || OS_BSD could then be converted to BASE_GTK.

Please add a description in build_config.h where you define it that explains the
difference between BASE_GTK and TOOLKIT_GTK.  BASE_GTK is for underlying system
features like clipboards and rectangles, whereas TOOLKIT_GTK is the UI toolkit
(since it's possible to have BASE_GTK and TOOLKIT_VIEWS).

http://codereview.chromium.org/118469/diff/1032/1033
File build/build_config.h (right):

http://codereview.chromium.org/118469/diff/1032/1033#newcode27
Line 27: #define OS_OPENBSD 1
This should probably be OS_BSD.  It would be nice to have this work across
free/net/open bsd.  You don't have to verify that it works on net/free bsd now,
when someone comes around wanting a freebsd port, they can fix it up then.

[pvalchev, 2009-06-11 23:15:22]

Thanks Tony, I was wondering about the cleanliness of the defines and other
BSD's too and your suggestion makes sense. There will be differences between the
three, so I suppose those can be handled with the standard #ifdef __OpenBSD__
inside the OS_BSD defines when the need arises (or we can define OS_XBSD and
then for each of them define OS_BSD as well)

We fixed the linker problem but I'm a bit too swamped this week to clean this
up, if I don't I'll clean it up next week - again, your suggested approach
sounds reasonable.

[tony, 2009-06-11 23:20:25]

Yes, that's right. I imagine all will define OS_BSD and each will also define
one of OS_FREEBSD, OS_NETBSD or OS_OPENBSD.  For now, we will put everything
under OS_BSD and let the next bsd port sort that out.

On 2009/06/11 23:15:22, pvalchev wrote:
> Thanks Tony, I was wondering about the cleanliness of the defines and other
> BSD's too and your suggestion makes sense. There will be differences between
the
> three, so I suppose those can be handled with the standard #ifdef __OpenBSD__
> inside the OS_BSD defines when the need arises (or we can define OS_XBSD and
> then for each of them define OS_BSD as well)
> 
> We fixed the linker problem but I'm a bit too swamped this week to clean this
> up, if I don't I'll clean it up next week - again, your suggested approach
> sounds reasonable.

[pvalchev_sightly.net, 2009-09-03 23:19:34]

I had started to clean it up with OS_BSD defines but then disappeared,
trying to resurrect this effort now and perhaps merge with Ben's
patch... just FYI.

On Thu, Jun 11, 2009 at 4:20 PM, <tony@chromium.org> wrote:
> Yes, that's right. I imagine all will define OS_BSD and each will also
> define one of OS_FREEBSD, OS_NETBSD or OS_OPENBSD. =A0For now, we will pu=
t
> everything under OS_BSD and let the next bsd port sort that out.
>
> On 2009/06/11 23:15:22, pvalchev wrote:
>>
>> Thanks Tony, I was wondering about the cleanliness of the defines and
>
> other
>>
>> BSD's too and your suggestion makes sense. There will be differences
>
> between the
>>
>> three, so I suppose those can be handled with the standard #ifdef
>
> __OpenBSD__
>>
>> inside the OS_BSD defines when the need arises (or we can define
>
> OS_XBSD and
>>
>> then for each of them define OS_BSD as well)
>
>> We fixed the linker problem but I'm a bit too swamped this week to
>
> clean this
>>
>> up, if I don't I'll clean it up next week - again, your suggested
>
> approach
>>
>> sounds reasonable.
>
>
>
> http://codereview.chromium.org/118469
>

[tony, 2009-09-03 23:24:29]

On 2009/09/03 23:19:34, pvalchev_sightly.net wrote:
> I had started to clean it up with OS_BSD defines but then disappeared,
> trying to resurrect this effort now and perhaps merge with Ben's
> patch... just FYI.

That sounds great, welcome back!

[pvalchev_sightly.net, 2009-09-04 16:25:47]

On Fri, Sep 4, 2009 at 3:06 AM, Ben Laurie<benl@chromium.org> wrote:
> On Fri, Sep 4, 2009 at 12:09 AM, <pvalchev@gmail.com> wrote:
>> Nice!
>>
>> So I had started a very similar patch for OpenBSD:
>> http://codereview.chromium.org/118469
>>
>> But I got really busy shortly after and went on vacation, I'd like to
>> resume & merge my patch too. The reason I didn't finish it is because I
>> wanted to clean it up and make a OS_BSD define like discussed in that
>> patch with tony@, matching all of the BSD's. Any chance we can do this?
>> I see you are adding OS_FREEBSD || OS_LINUX one-offs everywhere and the
>> same will have to be repeated for all of the BSD's... it would be nice
>> to make this cleaner.
>
> Sure, I'd be happy to go down this route. Presumably you'd still want
> there to be an OS_FREEBSD and OS_OPENBSD as well, for the, presumably
> few, places where they differ?

Great! And yeah, exactly, that's what I have in my tree:

So in build_config.h:
+#if defined(__OpenBSD__) || defined(__FreeBSD__) || defined(__NetBSD__)
+#define OS_BSD 1
+#endif

then for each individual one define OS_FREEBSD etc... too, as i'm sure
there'll be differences we need to account for.