[linux-dvb] Re: [RFC:PATCH] Build system improvement

[Trent Piepho]

On Thu, 22 Jun 2006, Mauro Carvalho Chehab wrote:
> Em Qua, 2006-06-21 =E0s 18:52 -0700, Trent Piepho escreveu:
> > The config-compat.h problem is still there. Unless you have more fixes =
for
> > this that aren't in Hg, I can fix this one.
> I did this fix on another test branch (just merging config-compat.h
> with .myconfig target), but I forgot to move this piece to the master.
> You can fix it if you want.

It is the script that makes config-compat.h that has the problem; the kerne=
l's
settings aren't overridden for options that are off.

> > Is there a reason you don't just read autoconf.h, which doesn't have an=
y
> > recursive includes to worry about?  It has all the config options in it=
=2E
> > The structure of config.h is somewhat more complex, for example the you
> > will read in asm/kgdb.h because config.h includes it.
> We may use autoconf.h, but when this were introduced? using config.h
> will take some more milisseconds, but it seems to be safer, in terms of
> backportability.

The autoconf.h file is in kernel 2.2.16 at least, in the same place and
appears to be the same format.  I didn't find it with kernel 2.0.36.
Somewhere in the 2.2 series should be plenty of back-compatability.

The thing I don't like about config.h, is that it is human written and
includes that kernel debugger header file.  These are not config format fil=
es
and so could have anything in them.  How can you be sure the file is parsed
correctly, when it could have anything in it?