Mail Archives: opendos/1997/02/03/17:11:15
At 20:27 1997/02/03 +0000, dg AT dcs DOT st-and DOT ac DOT uk wrote:
>>Have you folks thought about giving the inode number to 8.3 filenames?
>>An LFN app might see C:\Really.Long\Path.with.lots.of.dos\file23.tar.gz
>>and an 8.3 app would see C:i32456.dat.
>
[snip - example that basically says inode number is too hard to understand]
>I hate to say this, but the Win95 way: [snip - we all know it, right?]
>is probably the most intelligent way to go.
>
The only difficulty with the Win95 way is that the "~n" suffix that's
added depends on the order in which conflicting files are created. There-
fore, if we had two files "tutorial01.txt" and "tutorial02.txt" created in
that order, the 8.3 names would be a nice, logical "tutori~1.txt" and
"tutori~2.txt" However, if you then copied those two files to a different
directory, *in the reverse order*, the 8.3 filenames would be switched:
tutorial01.txt -> tutori~2.txt
tutorial02.txt -> tutori~1.txt
which would confuse the hell out of me, I can tell you.
Here at my workplace, we have a Linux/Windows network using Samba on the
Linux side. Linux filenames are served to the Windows machines with a
shortened filename that contains characters hashed out of the original
filename. It looks like this:
csear~a8.htm 1658 2-03-97 12:35 csearch1.html
csear~ts.htm 546 2-03-97 12:35 csearchc.html
csear~wg.htm 548 2-03-97 12:36 csearcht.html
csear~zk.htm 6514 2-03-97 12:35 csearch2.html
The names on the far right don't normally appear. They are the actual
Linux names.
I like this method because the short name is wholly and solely dependent
on the long name, meaning it's the same no matter where and when you
copy it. I don't know the function that's used to generate the 8.3 names,
but I'm sure there's a Linux guru out there somewhere who could tell us.
Just my $0.02,
Takashi
- Raw text -