Re: Fixing UMSDOS

Riley Williams (rhw@bigfoot.com)
Tue, 10 Nov 1998 12:26:53 +0000 (GMT)


Hi there.

>>> You are right that it should not assume that the files are 8.3,
>>> but the mangled names produced MUST be 8.3, otherwise you'll
>>> break every attempt to use UMSDOS on basic DOS machines.

>> I hope you don't mind my pointing out that under the VFAT file
>> system, EVERY FILE HAS AN 8.3 FILENAME. Furthermore, this is done
>> in a STABLE way...

> VFAT, IMHO, is a terrible kludge, which is in no way remotely
> compatibile with MSDOS FAT. The fact that sometimes you may access
> mangled VFAT names in MSDOS mode does not change that.

I agree that VFAT is a horrible kludge, but I would disagree with its
being incompatible with MSDOS FAT - in my opinion, it's far more
compatible with it than most of the so-called 'tools' supplied with
MSDOS itself.

If you want an example of this, try installing ANY DOS-based network
under ANY version of MSDOS from 3.20 through 6.22 inclusive, then run
such 'Standard' DOS tools as Norton's SpeeDisk (SD) program and watch
it mangle your entire filesystem in the process...due to the fact that
EVERY last one of those networks added extra fields to the directory
entry structure, always in undocumented ways, and usually totally
incompatible with any other type of network as well...and most added
extra hidden directory entries 'for accounting and security purposes',
as one network's technical documentation put it...

> Yes, UMSDOS is a kludge also, but (again IMHO) to a much lesser
> degree and also giving much more than VFAT (LFN is just one thing
> missing).

I can't fully agree here, although I agree with the general sentiment:
UMSDOS is far less of a kludge than most of those networks were, but I
don't personally think it's any less of a kludge than VFAT is...

>> ...something which the UMSDOS file system appears NOT to manage!!!

> I would like if you could elaborate on this.

OK...

> Yes, there are some inherit problems with UMSDOS. There are no true
> hardlinks, so they are emulated pseudo-hardlinks, and as such can
> be confused.

That's a problem which VFAT fails to even address, but not what I was
referring to...

> UMSDOS is tolerant to creating/deleting files/dirs in MSDOS mode
> (that is what umssync does), probably more than it is VFAT in MSDOS
> mode. It is completly transparent to standard MSDOS tools (chkdsk,
> directory sorters, etc) which is where VFAT in MSDOS mode utterly
> fails.

As stated above, the so-called standard MSDOS tools don't work with
most of the DOS-based netowrking protocols anyway, so one can hardly
regard them as any form of quality test, as you appear to imply...

> I would not say that "UMSDOS does not manage to be STABLE". In my
> tests, 2.0 seems stable enough, and current 2.1.x code is almost
> there. As a curent maintainer of that piece of code, I would like
> to hear reports of problems you had, and fix them (if possible).

The last time I tried to make serious use of UMSDOS was with kernel
2.0.33 so these comments may no longer apply. However, my experience
at that time indicated two bugs, both of which I consider major, and
both of which I reported. As far as I can tell, neither was ever
addressed:

1. UMSDOS was not producing correct short-form names for files with
more than one dot in them, where the ninth character of the
filename was a dot. I discovered this when it took an archive I
created under the name "MemAlpha.Systems.tar.gz" and mangled it
into "memalpha.sys" - and in the process overwrote an existing
file in the same directory!

2. In my opinion, it should respect the standard short forms for
compressed tarball extensions, but it fails to do so. For your
reference, I'm used to the following mappings:

*.tar.[Zz] => *.taz
*.tar.gz => *.tgz
*.tar.bz2 => *.tbz

From experience, it maps as follows:

*.tar.[Zz] => *.z
*.tar.gz => *.gz
*.tar.bz2 => *.bz2

Comments ???

>> Dare I suspect that you've never used the VFAT file system
>> yourself? Try the following test sometime:

> <test about copying file to VFAT and using it as MSDOS deleted>

>> Would you believe it, the file has a name that you can actually
>> use, even though MSDOS 6.22 and earlier do NOT support long
>> filenames...

> Do you want to say that you CAN NOT access UMSDOS file in MSDOS, or
> did you somehow deduced that poster above wanted to say that you
> cannot access VFAT files in MSDOS ? I see no explanation for any of
> those reasonings.

{Shrug} The author of the email I replied to emailed me privately to
apologise for claiming that one couldnae access VFAT filenames in
MSDOS in the said email, so I feel vindicated in interpreting his
comments that way...

> Opinions above are GNU-copylefted.

Chuckle...

Best wishes from Riley.

-
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/