Re: Kernel header changes break glibc build

[David Woodhouse <dwmw2 at infradead dot org>]

On Wed, 2006-12-06 at 15:23 +0100, Thomas Graf wrote:
> > I'm suggesting that if you want to change things around as you did, you
> > should make sure the users of those headers adapt to cope. You did fix
> > the in-kernel users; you neglected to fix glibc -- and as far as I can
> > tell you didn't even bother to _warn_ glibc folks.
> 
> I didn't warn them because I didn't know better. 

Fair enough. Now you do -- and thanks for fixing it.

> I was under the impression that glibc still maintains their own set of
> headers and will fix this automatically when they look at the diff.
> That's what I do for my userspace applications that use kernel
> headers.

That's never been the case for glibc. As I said, various people have
_attempted_ to do it, but it really isn't a workable approach and the
results were wildly inconsistent. Having a single, centrally-exported
set of headers is a massive improvement -- but yes, it does mean we have
to take a little care with what we're exporting to userspace.

> Ideally install_headers would do the trick but it often fails f.e.
> when some application which uses bsd features thus including net/if.h
> also wants to use new linux features and includes linux/if.h which
> then conflicts.

Applications in general shouldn't be doing that -- kernel headers are
for system libraries and tools only. We should try to ensure that the
_sensible_ use cases don't break, but we don't have to go overboard with
keeping our namespace clean and anticipating _every_ strange thing which
a userspace app might do with our headers -- we still get to say 'caveat
emptor'; it's just that we can't be _quite_ as haphazard about it as
we've been in the past.

-- 
dwmw2