Mail Archives: djgpp/2000/01/30/14:16:19
Hello.
This is a bit off-topic, but...
Eli Zaretskii wrote:
>
> Damian Yerrick wrote:
<snip>
> > If an entire long filename is uppercase, lowercase it before returning
> > it to the caller. That's what Linux vfat support does.
Unfortunately the Linux VFAT support isn't entirely like that. I have
experienced some weirdness with it. Here's how I think it happened (I may
be mistaken):
1. Install Internet Explorer 4 (or maybe 5, I haven't tried it) onto
Windows '95.
2. Wnable the option that allows you to type filenames actually as all
capitals (e.g. README.TXT) - it's in the Folder Options somewhere.
3. Reboot into Linux.
4. Look at the file README.TXT, see that Linux thinks it's README.TXT.
5. Create a file called readme.txt.
Now you have two files that technically have the same DOS/Windows
filename. Confused? I think what happens is that Windows '95 stores a long
filename, to get the correct capitalisation. I think Linux sees the short
filename as being distinct from the long filename.
This sort of thing has happened to me many times using Programmer's File
Editor, which seems to insist on saving all 8+3 filenames in capitals,
which causes the problem.
It has been said previously that the VFAT support in Linux is a hack. I
suffered quite bad FAT file system corruption with a stable kernel
(2.2.4?), so beware.
> Not only does the library already do so, it's even documented
<snip>
> Note that the library only does this for DOS 8+3 names; long file names
> are left intact, because otherwise names like GETTING.GNU.SOFTWARE would
> also be downcased. If Linux does that with long file names, Linux is
> wrong.
No, Linux leaves the long filenames as they are.
Bye,
--
Richard Dawe
richdawe AT bigfoot DOT com ICQ 47595498 http://www.bigfoot.com/~richdawe/
- Raw text -