intel driver will only compile with gcc

Hamish Moffatt hamish at cloud.net.au
Wed Jun 13 06:30:22 PDT 2007


On Wed, Jun 13, 2007 at 05:02:18AM -0400, Thomas Dickey wrote:
> On Wed, 13 Jun 2007, Donnie Berkholz wrote:
> 
> > Is there some definite date in the future when we can say, "It's now OK
> > to use features widespread in current compilers, if they have a reason
> > for use"? Or are we stuck with C89 forever?
> 
> you might have a point if the thread had started focusing on features of 
> c99 which weren't essentially cosmetic.  So far no one's made any valid 
> technical points past the feel-good/resume-enhancement argument (named 
> initializers ;-).
> 
> (Looking for _some_ technical discussion here, not finding any...)

Yes, Daniel gave some technical reasons in:
http://lists.freedesktop.org/archives/xorg/2007-June/025514.html

Specifically,

> They also have a great deal
> of value in partial initialisation.  The SI doesn't exactly have the
> smallest structures around, so having:
> foo = {
>     .name = "bar",
>     .other = thing,
>     .thing = meh,
> };
> 
> will definitely beat:
> foo = { NULL, NULL, 0, 0, NULL, 0, NULL, 0, 0, "bar", 0, 0, 0, 0, 0,
> NULL, NULL, NULL, NULL, NULL, NULL, thing, NULL, 0, 0, NULL, meh, NULL,
> NULL, NULL, NULL, NULL, 0, 0, 0, NULL, 0, NULL, NULL };
> 
> Bearing in mind also that the former is more portable towards adding
> struct members than the latter.

The linux kernel uses named initialisers extensively and I suggest it is
for this reason (partial initialisation).

Why could your favourite platform that wants to install Xorg 7.x not
also install gcc?

Hamish
-- 
Hamish Moffatt VK3SB <hamish at debian.org> <hamish at cloud.net.au>



More information about the xorg mailing list