[PATCH] system_x11 variant

[Jeremy Huddleston]

On Jan 4, 2009, at 01:17, Bryan Blackburn wrote:

> On Sat, Jan 03, 2009 at 10:28:48AM -0800, Jeremy Huddleston said:
>> Last week, I mentioned the idea of a system_x11 variant which would  
>> allow
>> users to continue using their ${x11prefix} X11SDK.  Below is a  
>> patch that
>> I'd like to get reviewed.  I've tested it, and it works here on  
>> Leopard.
>> I suspect xorg-libXcomposite (and others that aren't installed on  
>> Tiger)
>> probably need some additional pkgconfig-foo to find X11 (see the  
>> bottom of
>> xinit/Portfile if you're curious what I mean).
>>
>> This is essentially step 1 towards transitioning all ports to using
>> dependencies like "port:xorg-libXext port:xorg-libXaw" instead of
>> "lib:libX11:XFree86" or "lib:libXext.6:xorg-libXext lib:libXaw:xorg-
>> libXaw"
>>
>> This has the benefit of allowing us to develop against an identical  
>> X11
>> setup on all platforms while allowing expert users to set a variant  
>> if
>> they really don't want to use our X11 libraries.  Additionally,  
>> with this
>> variant set, we are better off that our current situation because  
>> we won't
>> install an X11 library into ${prefix} after a macports binary was  
>> already
>> linking against the one in ${x11prefix} (this is the main problem  
>> behind
>> http://trac.macports.org/ticket/17558 ).
>
> Two ports look odd: xorg-libs, which tests for libs.dylib (and being a
> meta-port, probably not meant at all here?);

Yeah, I noticed that after I sent the email.  That meta port should  
have no such variant.  Additionally xorg-xcb-util should have the  
variant but wasn't in the patch I submitted.

> and xorg-libpthread-stubs
> which tests for libpthread-stubs.dylib that I don't see in Xquartz
> (2.3.2.1).

Oh yeah.  Ditto for that.  That port isn't really needed for much of  
anything.

> Also, I'm curious about why you override the archive and activate  
> phases in
> these (as well as the X11 meta-ports xorg, xorg-apps, xorg-fonts,  
> xorg-libs,
> and xorg-proto).

Honestly, I didn't make xorg-proto, and I just used what was in xorg- 
proto as the base for xorg, xorg-apps, xorg-fonts, xorg-libs...

> Overriding activate causes them to be installed but then
> remain inactive, and hence show in 'port installed inactive' as well  
> as
> possiblity causing confusion for those who use 'port uninstall  
> inactive'.

Ok, so if I nuke those overrides, we should be good then, right?