Messages in this thread | | | From | "Robert de Bath" <> | Date | Fri, 10 Sep 1999 07:39:03 +0100 (BST) | Subject | Re: Your backup is broken! (W95 only :-) |
| |
On Sun, 5 Sep 1999, Roger Gammans wrote:
> Another I considered was to append ':ASH' to the file name and set the > case of the ASH bit appropriately.[1] Bad idea. Several things that change the names have been tried before, Prefix dots for hidden for example; they just don't work well. And as for putting the archive bit in the name, a filename should not change just because you touch it!
When you get down to it the S and H bits are a _lot_ like group ids in concept; how would you describe the S bit ? ... "A flag that marks the files that are special to the system" ... or perhaps "A field that marks the _group_ of files that are special to the system"
> The choice of colon as a separator was made as has little special > meaning in unix and is a banned character in FAT filenames. It's used as a Unicode seperator with VFAT.
> > Eyuck!! But ... > > The big problem with this is that it makes assumptions about the name > shortening algorithm used, AFAIK from my descriptions of VFAT the actual > method is not mandated (and certainly not enforced by the data > structures). So I would prefer a method which exposed a > vfat_set_short_name() [2] to userspace, but this is easier said then > done and has come up before. The point is this is the method used by Windows and it's a small enough subset of the possible methods to represent in this way. If you've got a file that has a strange shortname then it wasn't created by windows so even for windows it's shortname is probably irrelevent. But I don't really like the idea of putting the shortname in the gid anyway.
> Again a mount option to form filenames in a "SHRTNAME.EXT:TheLongName" > format might be one simple way round this problem, at least the idea of Yes I suggested this in a message with the subject "Your backup is broken" but while it's great for backups it's a royal pain to use plus you end up again with files that change name when you write to them. And as for symlinks; don't go there.
> both names being set together I _think_ alleviates some of the > complaints/problem with the vfat_set_short_name() method. 'vfat_set/get_short_name()' is a reasonable interface for saving shortnames that don't match a simple algorithm (eg number 1 or number 0 in my list)
> Additionally this mount option could be specified together with the > above mount option for the rest of FAT attributes. This is part of the problem; having the backup do a remount, or worse having the user remember to do it is very ugly indeed. If it can't be done (somebody is left logged in) your backup is ****ed. Much better to have a program that scans the disk for unusual files and records their oddness (SYSbits, ~87 shortname etc).
> > TTFN > > [1] Use Permission bit for Read_only, so this then exposes all FAT > attributes. > [2] ISTR, we already have a ioctl for vfat_get_short_name(). > -- > Roger Gammans > "If I have trouble installing Linux, something is wrong. Very wrong." > -- Linus Torvalds > > - > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in > the body of a message to majordomo@vger.rutgers.edu > Please read the FAQ at http://www.tux.org/lkml/ >
-- Rob. (Robert de Bath <http://poboxes.com/rdebath>) <rdebath @ poboxes.com> <http://www.cix.co.uk/~mayday>
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |