Mail Archives: opendos/1997/08/11/00:58:57
Rik wrote...
> i'm trying to keep my OpenDOS and Win95 partitions seperate, but i forgot
> to remove some long filenames from a couple files on the OpenDOS partition.
> now, when i try to delete the directory they used to be in (i deleted them,
> thinking it would eliminate the LFNs as well) i get an error saying the
> directory isn't empty.
This is annoying. If you use a disk editor to zero the bytes of the
directory where the longname parts of the now-deleted files reside, it
will probably not corrupt your disk :-}
I had a similar problem, at the stage of trying to re-install Win95.
It complained there was something wrong with the directory before
starting to properly install, and nothing other than a disk editor
fixed it. You might be luckier than me with a chkdsk though.
> i think that OpenDOS should totally avoid LFNs, at least in the crappy way
> Win95 does. One of my main reasons for hating Win95 is the crappy FAT
> system. as soon as Win98 comes out i'm switching over to FAT32 on that
> partition.
I agree that ext2 or something like that, done properly, would be the
best idea, by far. Remember that FAT32 doesn't really improve FAT
much, other than allow a smaller cluster size on large disks (the
downside is there is a *huge* amount of work to do in the File
Allocation Table, since each FAT will sprawl over many Kb, and it
retains the inefficient sequential list of clusters).
Chris wrote...
>... but the main problem with [ext2] would be compatibility issues...
One good thing is that the DOS call interface to LFN support (int 21h,
AH=71h) doesn't have to be tied to the MS way of implementing long names.
Software that uses that API could be made to work with ext2fs, hpfs, ntfs
or other systems. Programs should not normally try to access the disk
sectors directly to "do" long names, but go through the system.
What *is* a problem, is where there is a mixture of old an new
software; it has been suggested that "legacy" programs be excluded from
access to any long-named files to avoid risks (OS/2 pretty much does
this). There might be a safe middle ground, but it is a nighmare (under
OpenDOS or WIN95) when you consider the number of ways the present (MS)
design of the LFN system can go wrong.
-------------------------------------------------------------------------------
Mark Aitchison, Physics & Astronomy \_ Phone : +64 3 3642-947 a.h. 3371-225
University of Canterbury, </ Fax : +64 3 3642-469 or 3642-999
Christchurch, New Zealand. /) E-mail: phys169 AT csc DOT canterbury DOT ac DOT nz
#include <disclaimer.std> (/' Callsign: ZL3TQE
-------------------------------------------------------------------------------
- Raw text -