> This is like the Stanford checker stuff. These are bugs. You have
> the means to find them automatically, but not the time or desire to
> fix them.
Actually I have got the desire to fix them, what I lack is the ability
to get patches into the kernel that are too non-trivial to go through Rusty.
> Post a list and perhaps others will pitch in. Something
> drivers/ide/Config.in:17: In dep_tristate CONFIG_BLK_DEV_IDESCSI:
> drivers/ide/Config.in:17: CONFIG_SCSI not defined for i386, ppc, ...
Ok, but perhaps it's not clear how many problems there are. The full
log file from a gcml2 run on 2.5.29 is 573 KiB. Here's a summary:
28 unset-statement (ignore these)
These numbers are aggregates over 17 arches, so you need to divide by
a number between 1 and 17. Also some of these have been fixed in 2.5.30.
I can post the full list if people want, but it would be a bit overwhelming.
> In this case the fix would be to move CONFIG_BLK_DEV_IDESCSI to the
> SCSI subsystem, where in my opinion it belonged anyway.
> This would break down if there are any actual cycles - things which
> can't logically be moved to a place after the definitions of the
> facilities on which they depend.
I'm not able to detect anything like that.
> That we have to worry about this at all is an artifact of using a
> procedural langauge, rather than a declarative language like Prolog or
Actually config language isn't really procedural, pseudo-procedural would
be more like it.
> I *really* don't want to go *there*, though. (:
> > And "" != "n" in other contexts, like if ;then statements.
> That is true. But that should not affect the dep_* logic, should it?
Correct, I'm just pointing out that using orthogonality arguments with
the config language is not going to get you anywhere useful.
> The point here is that people who aren't hip-deep in config language
> code don't think about them being separate. Ergo bugs.
> I've been thinking hard about a new sort of 'if' statement that
> doesn't look like a test command.
Interesting, I'd like to hear more.
My favourite idea is a different form of choice which worked more
like a menu, so you could intersperse conditionals with the choice
items. This would help with the several places in CML1 were the
same choice is mostly-duplicated in different conditionals, e.g.
'Kernel page size' in arch/ia64/config.in. ESR worked out a sensible
set of semantics for how this should work.
> (Yes, it's my mission to eventually
> get rid of the '$'s in config language. I think it can be done,
> piecemeal, over a long period of time.
Sounds good to me.
> This is if we don't end up
> adopting a whole new language like CML2 or Roman's stuff.)
Or a new parser & frontends for the existing language.
I'm not convinced that a complete new language will ever succeed after
the CML2 debacle, machine translated or not. This is why I gave up the
idea of automatically converting CML1 to CML2.
-- the price of civilisation today is a courageous willingness to prevail, with force, if necessary, against whatever vicious and uncomprehending enemies try to strike it down. - Roger Sandall, The Age, 28Sep2001. - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to firstname.lastname@example.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/