Re: bugfix, hostfs

[Tomas Ebenlendr]

> >> This requires some changes in the cache manager, but not difficult. What 
> >> do you think?
> >
> > Hmm, leaving this change for anybody who understand disks in grub.
> > Instead of it I use disk->name :-) .
> 
> Better use an int.  It is easier/cheaper to check and IMHO cleaner.
> 

Hmm, I check it via macro GRUB_FAKEDISK_IS_HOST(disk), and I want to
know it is 'host' fakedisk, because then it is not problem to add other
fakedisks. The best way is probably Okuji's proposal, but I want it as
single patch. (I'll probably do it next week.)

As worse problem I see utf8 from grub side and 8-bit encoding from libc side
(pathnames). This can be solved with some translation, but I'm not
decided how to tell grub the encoding.
One way is to have device.map entry style:
host    iso8859-2
This will also specify that 'host' device will be accessible only if
defined in device.map.

> > +      case ENOMEM: r = GRUB_ERR_OUT_OF_MEMORY; break;
> > +      case ENOTDIR:
> > +      case ENOENT: r = GRUB_ERR_FILE_NOT_FOUND; break;
> > +      case ELOOP: r = GRUB_ERR_SYMLINK_LOOP; break;
> > +      case EIO: r = GRUB_ERR_FILE_READ_ERROR; break;
> > +      default: r = GRUB_ERR_BAD_ARGUMENT; break;
> 
> I think I missed this last time.  Better use multiple lines for a case
> statement.
>

Thanks. That is relict from '"nice" coding standard' which we used for our
school project. ("nice" coding standard says: do it the "nicer" way.). I
will fix all GCS errors that you found.

Btw.: does anybody here use vim? Is there GCS-syntax file for vim?
-- 
                                 Tomas 'ebi' Ebenlendr
                                 http://get.to/ebik
                                 PF 2004.61865285367