Rockbox mail archiveSubject: Re: FAT Filename Mangling bug. (was: RE : Regarding the recent recording PANIC's)
Re: FAT Filename Mangling bug. (was: RE : Regarding the recent recording PANIC's)
From: Michael O'Quinn <michael_at_oquinn.info>
Date: Thu, 24 Apr 2003 13:42:12 -0700 (PDT)
On Thu, 24 Apr 2003, Björn Stenberg wrote:
> Michael O'Quinn wrote:
> > Not true.
> > There are specific mapping rules that are defined for mangling long names
> > into short. And if it's FAT (of whatever flavor) the short names are
> > THERE, even if the OS normally hides them
> > Here's what Micro$oft says in the Windoze-98 Resource Kit...
> Yay, a docs arms race! :-) I'll raise you one:
Ooooh! How fun! And, since we are across the Pond from each other, we'll
have to use ICBD's, that is, Inter Continental Ballistic Documents...
> This document, "FAT32 File System Specification", states as the only
> requirement that "the short name is automatically generated from the
> long name in such a way that it does not collide with an existing short
It may be that the reason it's choking is something OTHER than the name
not matching. It could be something with the association between the
various directory entries, or the checksum, or some undocumented
feature. (Not that THAT would EVER happen with such a large and
professional outfit.) I'll poke around with a disk editor and see if
anything obvious jumps out at me.
> > Long File Name Support in Windows 98
> That document is a description how Windows 98 creates short filenames,
> not a fat32 filesystem specification.
And we ALL know how well Microsoft documents the ACTUAL behavior of their
OS's in their published "Official" specs... ;-)
> > But this does affect the ability to open the file from WIndoes 98, and it
> > also confuses JBSync when it sees a new playlist created by RockBox.
> Interesting. We've done it this way since november and you are the first
> to even notice it.
> > it is worthy of it's own bug report.
> Only if you can convince me that I am actually violating the
> specification and not just doing it differently than the win98 code.
The heck with the "Official" spec. We all know that Microsoft has a long
history of doing undocumented and incompatible things that give other
developers histrionics. It DOESN'T work on my Windows boxes, and Windows
finds a definite error. THAT is the problem.
Let's see now, Rocket Engines, Nose Cones...
AH! There we are...
I really *will* look at this more, particularly at the directory entires
themselves. And I'll post the text of ScanDisk's complaint. I honestly
don't know why I'm the first to notice this. Maybe... I just don't know.
Has anyone else seen this behavior?
Received on 2003-04-24