Utility to check if XP file names exceed 255 chars

[Renan]

void Goeroeboeroe:

I've quite a lot of applications, so my Startmenu is pretty deep
organized. So I click for example on Start -> All Programs -> -Internet
-> Mail and Newsgroups -> Utilities -> .... -> Application
When this all together exceeds 255 characters, I get the BSOD, even if I
only hover over the name of the application.
It shouldn't be too hard to duplicate it, but maybe it has something to
do with my set-up, I don't know.

I tried recreating it with a Windows XP installation in VMware:

Created various folders, one within each other, named
'12345678901234567890', so I would know each filename has 20 chars.

And, at a certain point, Windows told me: "Unable to create folder, file
name too long" (or similar words). I was unable to create any more
directories.

I mounted the directory though a network share and was able to create
larger names, but what I got when I accessed the file, was just "Access
denied".

No BSOD here, but Windows freaked when I tried to delete the folder.

What is your filesystem? FAT32 or NTFS?

 

[Intermagic]

When this all together exceeds 255 characters, I get the BSOD, even if I
only hover over the name of the application.
It shouldn't be too hard to duplicate it, but maybe it has something to
do with my set-up, I don't know.

96% of the BSOD are DRIVERS problems. Of course, regular people and newbies
blame Windows for this , but this is a driver problem.

 

[Goeroeboeroe]

void Goeroeboeroe:


I tried recreating it with a Windows XP installation in VMware:

Created various folders, one within each other, named
'12345678901234567890', so I would know each filename has 20 chars.

And, at a certain point, Windows told me: "Unable to create folder, file
name too long" (or similar words). I was unable to create any more
directories.

I mounted the directory though a network share and was able to create
larger names, but what I got when I accessed the file, was just "Access
denied".

No BSOD here, but Windows freaked when I tried to delete the folder.

What is your filesystem? FAT32 or NTFS?

NTFS. But it can have something to do with something odd in my comp.
Still I think Windows should prevent this.

 

[Goeroeboeroe]

96% of the BSOD are DRIVERS problems. Of course, regular people and newbies
blame Windows for this , but this is a driver problem.

Driver or not, I think Windows should prevent this. And thanks for the
compliment: it must be twenty years ago somebody called me regular or
newbie )
Since you know this is a driver problem: could you explain this? How do
you know this? Which driver is the cullprit? I'd be really thankful if
you could tell me this, because all my watchdog-applications couldn't
find it. But of course, I'm just regular or a newbie...

 

[Ben]

I'd like to see someone make a filename+path utility where you can specify
the length of the filename and a checkbox if you want the path included
also. You should be able to enter any number you wish: 8+3, 28+3, 61+3, 255.
This should be a box where you type data, not a dropdown list where you only
have a few choices. Some webforms have character limits for certain fields.
In such case, you might enter 40. This would be helpful for preparing
backups to burn CD/DVD, where the limit is 61+3, and I believe that means
filename+path.

I have now rewritten my original DOS / gawk script as a simple VBS which
suits my needs, but I'm going to have a crack at converting it to VB and
adding a simple GUI to specify directories, options for length of path /
filename etc rather than them being hardcoded.

Does anyone know whether the backslashes which will be found in a fully
qualified path should be counted towards the limit?

i.e If you had 125 single letter directory names under each other,
\a\b\c\d\e etc, would you have a path that counted as 250 chars, or 125?

It seems like a silly question, but it did occur to me that if the \'s
don't count, my original script would overestimate lengths since it just
took the length of a full path rather than ignoring the \'s.

B

 

[Intermagic]

Driver or not, I think Windows should prevent this. And thanks for the
compliment: it must be twenty years ago somebody called me regular or
newbie )

Well, pray tell me, how in the world can a OS prevent a driver crash? A
driver works in KERNEL mode!!!

 

[Marten Kemp]

Well, pray tell me, how in the world can a OS prevent a driver crash? A
driver works in KERNEL mode!

A properly constructed OS can't be brought down
by anything short of a hardware error. Bounds
checking, reasonable-value verification,
ruthless testing by people who didn't write
the software -- OS writing should be an
exercise in paranoia.

Drivers needn't run in kernel mode -- modular,
mini-kernel OSes run the drivers in user mode,
sacrificing a small amount of performance to
gain a large measure of crash-resistance.

--
-- Marten Kemp
(Fix name and ISP to reply)
-=-=-
.... "I don't think so," said Rene Descartes. Just then, he vanished.
-tftd, Nov 13/02
* TagZilla 0.059 * http://tagzilla.mozdev.org

 

[Intermagic]

Drivers needn't run in kernel mode -- modular,
mini-kernel OSes run the drivers in user mode,
sacrificing a small amount of performance to
gain a large measure of crash-resistance.

Well, tell me one single OS which cannot be taken down by a bad video driver
(for example).

Have you never seen a Kernel Panic Error (Same than BSOD) in KDE with bad
Ati drivers? Take a look at the newsgroups and you'll see. Or a Panic error
in BeOS with Crystal sound drivers?

The is NO protection for bad drivers. A driver is a driver is a driver.

 

[Marten Kemp]

Well, tell me one single OS which cannot be taken down by a bad video driver
(for example).

Have you never seen a Kernel Panic Error (Same than BSOD) in KDE with bad
Ati drivers? Take a look at the newsgroups and you'll see. Or a Panic error
in BeOS with Crystal sound drivers?

The is NO protection for bad drivers. A driver is a driver is a driver.

And you know what they say about the nut behind the wheel <grin>.

You have too narrow a definition of "OS" -- limiting yourself to
the ones that run on desktop-type machines. The OSes that run
industrial-scale hardware (zVM, zOS, MVS, OS/400 and the like)
don't have those vunerabilities.

My favorite OS, zVM, runs a lot of processes (TCP/IP, for example)
in their own virtual machines, totally isolated from the OS itself
and all other running processes. If such a process fails, however
spectacularly, it's not going to bring down the rest of the system.

There *are* places where the code is insufficiently paranoid but
they're extremely few and generally cause problems with broken
hardware. The only one of which I'm aware had to do with an ancient
disk device in a testing center (so broken that the platters
weren't spinning any more and so old that IBM no longer had the
parts to repair it; it hadn't been thrown in the trash because
it was being used to daisy-chain another ancient device that still
worked enough for testing purposes) that returned a zero for its
capacity, so the equivalent of a 'df' command wound up generating
a divide-by-zero error down within the kernel.

Aside from that, there was one incident where a system went down
because of a catastrophic hardware failure of a swap-space volume
(they woke me at 0300 for that, of course -- it took about 30
minutes to isolate the failed volume and get the system back up),
and another when one processor (out of 12) failed and the system
hadn't been configured for fault-tolerance (it was a low-serial-
number system that hadn't had all the hardware fixes applied).

So, while no OS is bulletproof some OSes are proof against
larger calibers than others.

--
-- Marten Kemp
(Fix name and ISP to reply)
-=-=-
.... To sysadmin or not to sysadmin... that is the question, whether tis
nobler
in the minde to suffer the slings and arrowes of outragious fortune, or
climb to the top of the building with a ****ing high-power rifle and scope.
--Greg "Twotone" Spiegelberg
* TagZilla 0.059 * http://tagzilla.mozdev.org

 

[Intermagic]

You have too narrow a definition of "OS" -- limiting yourself to
the ones that run on desktop-type machines. The OSes that run
industrial-scale hardware (zVM, zOS, MVS, OS/400 and the like)
don't have those vunerabilities.

Excuse me but thos OS are running in special designed hardware which only
have, say 12 different TESTED drivers. Compare that to the MILLIONS of
different drivers that desktop OS must run on.... This is like comparing
Pong to Flight Simulator.

 

[Marten Kemp]

Excuse me but thos OS are running in special designed hardware which only
have, say 12 different TESTED drivers. Compare that to the MILLIONS of
different drivers that desktop OS must run on.... This is like comparing
Pong to Flight Simulator.

Yes, that's true. And you might have more precisely
defined which OSes about which you were making such
sweeping statements <grin>. That the design of the
OSes is such that a driver failure can bring down
the entire system is a comment on the design of
the OS, don't you think?

I'm not that conversant with the driver structure
of e.g. Linux, but wouldn't it be possible to run
the client part of X-windows as a kernel process
and the server part (that deals with the actual
keyboard, mouse and display hardware) as a user
process?

The Minix version 3 project hasn't gotten X
ported (AFAIK), but Minix is a microkernel that
runs such things as file and memory management
as user processes, for exactly the reliability
reasons I stated.

--
-- Marten Kemp
(Fix name and ISP to reply)
-=-=-
.... "Why does Sea World have a seafood restaurant?? I'm
halfway through my fish burger and I realize, oh my
God.... I could be eating a slow learner."
--Lynda Montgomery
* TagZilla 0.059 * http://tagzilla.mozdev.org

 

[Goeroeboeroe]

Well, pray tell me, how in the world can a OS prevent a driver crash? A
driver works in KERNEL mode!!!

Been away, do I'm a bit late with my answer.
The OS could prevent it if it checked for not exceeding 255 characters.
And, with all respect, I don't see what a driver has to do with just
hovering over a name. I didn't click or anything else like that, just
hovering over it.

 

[Goeroeboeroe]

(e-mail address removed) says...

NTFS. But it can have something to do with something odd in my comp.
Still I think Windows should prevent this.

You made me curious. I tried again to replicate the error, but now I get
the same as you: name too long. I've reinstalled Windows some months
ago, so it probably had something to do with my old configuration. Or a
damaged file, who knows...

 

[Luis Cobian]

Been away, do I'm a bit late with my answer.
The OS could prevent it if it checked for not exceeding 255 characters.
And, with all respect, I don't see what a driver has to do with just
hovering over a name. I didn't click or anything else like that, just
hovering over it.

The OS does **check*** the length of the name. try it yourself. The poster
that say that this is a driver problem seems to be right. Almost every BSOD
is a driver's foult in my experience.

 

[Luis Cobian]

If path and name in Start -> All Programs -> ... exceed 255 characters,
it even leads to a BSOD if you hover over the name.


This is what makes me think that this is a video driver problem.
Hovering implies a lot of activity in the graphic card. There are a lot
of problems related to hovering with some cards (disapearing icons in
some palets, corrupted images, etc). Older ATI drivers were horrible
handling this. Fortunatly later versions seem to have fixed those
problems.

By the way,XP can handle paths > 255 characters (more than 35000 IIRC)
but only in unicode mode. Because must of applications out there are
not unicode enable, the OS rightly warns about the 255 limit, avoding
this way that thousands of applications break while trying to access
those files.

 

[Goeroeboeroe]

If path and name in Start -> All Programs -> ... exceed 255 characters,
it even leads to a BSOD if you hover over the name.


This is what makes me think that this is a video driver problem.
Hovering implies a lot of activity in the graphic card. There are a lot
of problems related to hovering with some cards (disapearing icons in
some palets, corrupted images, etc). Older ATI drivers were horrible
handling this. Fortunatly later versions seem to have fixed those
problems.

By the way,XP can handle paths > 255 characters (more than 35000 IIRC)
but only in unicode mode. Because must of applications out there are
not unicode enable, the OS rightly warns about the 255 limit, avoding
this way that thousands of applications break while trying to access
those files.

Okay, this makes sense. After reinstalling Windows some time ago I
installed another video driver, and the problem isn't there anymore.
But it still doesn't explain everything. I was able to make such a long
path in the first place, without any warning. And I think that's a job
for the OS to do.
Problem is that you have all those brilliant monitoring programs, but
with a real crash they don't (can't) give any information. Same for
event log etc.