Windows 95 support chap skipped a step and sent user into Micro-hell

Greetings, gentle reader, and welcome once again to Who, Me? in which Reg readers like yourself try to make each Monday a little less manic by sharing tales of foible and fallibility. Take, for instance, this week's hero, who we will Regomize as "Bill". Many, many years ago Bill toiled in a call center where he offered tech …

[Spazturtle]

Still one of the best ways to fix Windows.

I recently had Windows refuse to boot and it was giving all sorts of errors about how the boot information is corrupt. Now this should be an east fix with a Windows install USB and running "Bootrec /RebuildBcd" or various other similar commands in the command prompt, but none of that was working as it was saying I didn't have permissions, which is odd as the installer should be running as admin.

Then some guy in the comments of the microsoft support forum said to ignore all that and just tell the installer to blind dump the boot files at the start of the disk, and gave a command to do so. Boom, Windows booted back up.

[corestore]

Bogus

I'm surprised such an entirely bogus story slipped through El Reg's expert hands!

First, deltree does not 'delete the windows system'; it will delete ANY subdirectory tree you *tell it to delete*.

Second, chkdsk doesn't mystically identify 'important data' you may want to keep; rather, it checks the entire filesystem *integrity*, and identifies and may try to repair any *corrupted or truncated* files. It's the DOS equivalent of fsck really.

[Philip Storry]

Re: Bogus

I'm willing to chalk that up to a slip on the author's part.

We have two main issues - the first is that the person telling it isn't a journalist so may not be phrasing things well, and Matthew JC Powell has to deal with that.

The second is that this was somewhere between 29 to 24 years ago, and therefore the memory may be a touch hazy.

It's good to have high standards, but I find these minor mistakes fairly easy to forgive in these circumstances.

[FIA]

Re: Bogus

C:\> chkstry /y

Checking story on R:

The type of sory is HISTORIC

CHKSTRY is verifying overall integrity (Stage 1)...

Found basic structure.

Integrity verification complete.

CHKSTRY is verifying details (Stage 2)...

Possible inconsistent use of DELTREE.

Possible inconsistent use of CHKDSK.

CHKSTRY is verifying indexes (Stage 5)...

CHKSTRY has scanned the story and found minor issues.

You may continue to enjoy The Register however you should run

CHKSTRY with the /f option to fix any errors found.

3373 Total characters

625 Total words

21 Total paragraphs

C:\>

[KittenHuffer]

Re: Bogus

But today you can fix those issues using Regedit!

-----------> Mine's the one with Computer\HKEY_LOCAL_MACHINE\SYSTEM\RNG\Seed set to 0!

[TeeCee]

Re: Bogus

I see you are trying to take the piss. Would you like help with that?

(Where's the sodding paperclip icon when you need one?)

[simonlb]

Re: Bogus

From memory (and a quick Google), deltree was released in MS-DOS 6 and retained through Win 9x. But yes, a chkdsk first would definitely be in order.

[Anonymous Cowpilot]

Re: Bogus

Something similar to what is described could be a misremembered scenario. As you say, deltree deletes everything under the specific directory so deltree c:\windows would do a reasonable job of removing windows. However as a dos utility it wouldn't follow windows shortcuts and would instead delete them, and windows didnt support real links so I can't see a scenario where it would escape from the windows directory and delete the whole disk. Chckdisk would help clean up lost sectors and mismatches between the FAT and what is on the disk, but would not do anything clever with directory trees.

Most likely to cause the described behaviours would be if the person ran deltree c:\ instead of c:\windows

[heyrick]

Re: Bogus

Just going by the story, it's possible that DELTREE escaped from \Windows and nuked stuff elsewhere because, as the story says, crosslinked files. This isn't a symlink, this is where the FAT directory information is messed up. CHKDSK would have noticed and "fixed" it to a degree...though if the drive structure is so completely fucked that DELTREE in \Windows wipes everything else, formatting and starting again is probably the best solution (I can imagine many broken files).

[katrinab]

Re: Bogus

FAT16 and FAT32 (Win95 OSR2) don't support links. NTFS supports junctions which are basically the same thing, but Windows 95 can't read NTFS volumes.

Crosslinked files happens when you have 2 or more different FAT entries pointing to the same cluster on the disk, which shouldn't happen, and means your data is corrupted and almost certainly lost anyway.

[John Brown (no body)]

Re: Bogus

Injudicious use of any/all of join, append, assign, subst can cause interesting results when following a directory tree.

[Prst. V.Jeltz]

Re: Bogus

First, deltree does not 'delete the windows system'; it will delete ANY subdirectory tree you *tell it to delete*.

Yes , thats not in debate , the author was telling you which directory he was using it to delete !

wether deltree would follow a shortcut or sytem link out of that tree is another question

Also how deltree would do this with the files in use is dubious.

does that mean he talked the user , on the phone , through booting on a diagnostic disk , before scraping windows off the disk and putting it back ?

and then all the other crap that goes with that?

That is a worst nightmare scenario

[Anonymous Coward]

Re: Bogus

Given that the OS in question was Windows 95, which had FAT32 as its underlying filesystem, then symlinks are out of the question. Deltree considered shortcuts to be a regular file and didn't follow them.

[Spazturtle]

Re: Bogus

Symlinks are out of the question, but hard link can happen (no OS will let you willingly create them but corruption can cause them). Chkdsk will try and fix any hardlinks it finds.

[jfm]

Re: Bogus

Bit late to this but ln(1) exists to create hard links (by default, or symlinks with the -s flag) on UNIX-like systems, so unless I'm misunderstanding you, "no OS" is a bit sweeping.

[Sandtitz]

Re: Bogus

"Given that the OS in question was Windows 95, which had FAT32 as its underlying filesystem"

FAT32 support came out with Windows 95 OSR2, so it could have been FAT16 as well.

[Lee D]

Re: Bogus

Okay, personal history:

A friend was on our Windows 3.1 machine and was helping us clear out files.

They deleted a perfectly innocent folder.

However, because of the exact-same "cross-linked chains", we noticed that the file deletion was taking a while - and was in fact looping around to the C:\ drive root despite having started inside a deep subfolder. Cross-linked chains literally means this - one of the folders was "linked" to have a sub-folder which was actually ABOVE it in the directory tree.

When we walked over to the machine, we immediately spotted filenames like C:\Temp\harmless\folders\here\windows\system\important.dll. And before long, C:\Temp\harmless\folders\here\harmless\folders\here\windows\system\important.dll and so on. Unfortunately those filenames were whizzing past on a Windows delete dialog!

We screamed CANCEL IT! and then investigated the state of the OS - those files were indeed missing from C:\windows\system along with pretty much the entire rest of the filesystem. We were more impressed that Windows kept running than anything else.

Fortunately, we were able to do two things - first, reinstate AUTOEXEC.BAT and CONFIG.SYS, and second find the Windows installer disks. We had to reinstall Windows, but our programs managed to survive and we were able to take a copy of the registry (I can't even remember how, to be honest, but those files were intact - and back in those days "in use" files weren't as locked as you would expect them to be).

We rebooted into DOS, ran chkdsk (because at this point we didn't know what was happening) and it found thousands of cross-linked chains that it offered to resolve. That took ages but did, indeed, resolve the circular filesystem tree problem, and then we were able to recover what was missing and reinstall the rest.

So... yes, deltree will delete the windows system on old Windows, if you have cross-linked chains (as the article asserts!). And chkdsk is able to resolve cross-linked chains through its filesystem integrity checks.

This was in the days of FAT, etc. but I don't think it's possible nowadays.

[Jou (Mxyzptlk)]

Re: Bogus

> This was in the days of FAT, etc. but I don't think it's possible nowadays.

Yes and no.

Yes, because what once was called cross linked files can now be used to save space. NTFS offers hardlinks for this. Since Vista this trick is used to save quite some disk space. Get "Hardlink Shell Extenstion", and then enter Windows\System32. In princpal even on the part-match file contents can be "cross-linked", but on purpose (reparse points as example). And NTFS is not the only Filesystem able to do such thinks.

No 'cause it is not seen as an error.

[Mtrain4004]

Re: Bogus

It would be scandisk, not chkdsk. Chkdsk is NT based

[Lee D]

Re: Bogus

Nope. CHKDSK.

"The command is available in MS-DOS versions 1 and later."

CHKDSK interrogrates filesystem structure.

SCANDISK tested the physical "holding" of data on each individual sector through read and read/write tests.

They are two different commands for two entirely different functions.

[Jou (Mxyzptlk)]

Re: Bogus

"entirely different functions" ? So Scandisk never ever checks the filesystem structure?

Go back to DOS 6.22, and then verify yourself. You can use "help chkdsk" and "help scandisk", since this version of DOS has the manual included electronically.

Or read the wikipedia information.

[Lee D]

Re: Bogus

Scandisk: "The program was first introduced in MS-DOS 6.2". Up until then, you could only use chkdsk. Both programs evolved over time and pretty much merged together later, but at the time they were entirely distinct, and you didn't surface-scan with chkdsk or filesystem check with scandisk. Early scandisk didn't work well on Windows-based systems, chkdsk could be run in a command line window effectively.

https://www.computerhope.com/chkdskh.htm

https://www.computerhope.com/scandisk.htm

"You cannot run ScanDisk from a command line while Windows is running"

Also note that pre-Vista chkdsk doesn't have such a warning, and had extremely limited options and so wasn't a scandisk replacement at the time (but would later become one).

It was designed as a successor, but there's a reason that CHKDSK (and later autochk) are still present even in modern Windows in various forms - and scandisk later died. CHKDSK of this time doesn't perform any surface scan (and still doesn't?), SCANDISK did. They were entirely different programs throughout their entire history of being directly executed by users, rather than forming part of an automated boot-time filesystem check in modern Windows.

(And not least: "However, ScanDisk cannot check NTFS disk drives, and therefore it is unavailable for computers that may be running NT based (including Windows 2000, Windows XP, etc.) versions of Windows; for the purpose, a newer CHKDSK is provided instead.")

[Jou (Mxyzptlk)]

Re: Bogus

> you didn't surface-scan with chkdsk or filesystem check with scandisk

> https://www.computerhope.com/scandisk.htm

If you actually read the link you provided you will see that scandisk DOES a filesystem check & repair. Not only that, it shows the "MS-DOS 6.0 and earlier syntax" options too. What is wrong with you?

[Anonymous Coward]

"the DOS equivalent of fsck"

Hmm.

Is that command another (but uncaught) Easter egg in this vein?

[Philip Storry]

Scandisk?

Having worked in a call centre near Sidcup* that handled Microsoft's newly outsourced OS support in 1995, I have one question...

Why not scandisk? Easier to use, does most things better, and has one huge advantage... the surface scan.

The integrity check of the filesystem's structure will take a couple of minutes, but then you're into a nice long scanning operation.

Which gives you the excuse to wait five minutes for them to realise that this will take forever, and then tell them to call back if it encounters issues. Allowing you to go home on time.

See? It may be almost thirty years ago, but I still have my support skills. ;-)

---

* Any other FCY03 survivors out there? FCY01 is a housing association office, 02 and 03 are now flats. The Seven Stars is still open, and I'm tempted to swing by one day for old time's sake...

[herman]

Re: Scandisk?

Eeeevullll…

[simonlb]

Windows can also break Windows

A long time ago when I did 1st line support I was remotely removing some old profiles on a Windows 2000 workstation to free up disk space for the user when the machine decided to also delete the profile of that same logged in user. I hadn't selected that profile for deletion but all the desktop icons disappeared before an error popped up saying some files were locked. After a quick check to confirm this I called over my supervisor who took one look then said, "What the f..?"

We passed the call over to the technical team who were eventually able to recover any of the users data which had been deleted, but they confirmed that I hadn't done anything wrong but that they suspected there was some corruption within one of the profiles which had caused the issue. They'd also never seen this happen before.

[Jou (Mxyzptlk)]

Re: Windows can also break Windows

A possible source of this: Junction/softlink/hardlink (not to mix up with .lnk files). But I don't have a W2K machine at hand to check whether something like that is used in Windows 2000 by default. But even if not, a user can create those manually pointing to a higher level directory, and Windows Explorer might follow it for the recursive delete.

[ColinPa]

Don't follow the instructions

One of my first jobs ( 40+ years ago) was working with DOS/VS running under VM on the mainframe. You had one disk for the system, and one disk for the user data. My job was to recompile everything and put it on the user disk. We had two user disks, and would flip/flip between build and other people using it.

Day 1, I followed the instructions, erased all the files on the user disk and started building.

Half an hour later someone wandered round saying is there a problem with...

The instructions had said link in disk xyx... so I did, and wiped out the current disk.

My supervisor said (face palm moment) that is an example of the command - not the actual command.

The documentation was changed to make it clearer

Day 2, I followed the new instructions.

Half an hour later people came round and said... is there a problem with....

The problem was that an old disk was "lying around" and the system used that in preference to the one I was trying to use.

The documentation was then changed to 1) remove all user disks from the configuration, 2) link in disk

Day 3 I was moved to a different project.

[that one in the corner]

Re: Don't follow the instructions

Sounds like you had the perfect start for an illustrious job on the Tiger Team.

[MarthaFarqhar]

Re: Don't follow the instructions

Why bother with documentations, manuals and instructions. The probability of them being a) absolutely clear and correct and b) followed to the letter tends to 0 the more complex the task.

[Alistair]

Re: Don't follow the instructions

I've written documentation for processes that admins had to follow. In order to get things down clearly for people we knew *damn* well would not be fully aware of what they were doing when executing said documented processes. In one set of documents we were covering migration of (new version of application) from final QA approval system to (staging area on production system). This, due to the length of time it took to execute the copy and setup had to be done during *sigh* production hours. Since we were aware of all the of the actions there were not one, two but THREE checks the user was to execute before firing up the command chmod -r 750 /path/to/staging/files. All three of those checks were to ensure that a) the user's shell was in the correct directory b) the user was on the correct volume and c) the user could see the correct (previous staging kit).

This set of documentation had worked well for 6 previous releases. New dude in outsource corp. Decides that the checks aren't needed. Hell, the PATH isn't needed.

Fortunately, once up and running on *nix systems, permissions on libraries and code on disk aren't much of an issue, but the "I can't log in" calls were wild. Try chmod 750 on /usr/lib/* on HPUX. Fun times. Fortunately I had an open shell on the system at the time. And, two other active systems that didn't get hit to compare against. Took me most of an hour and a half, but the fun thing, No outage to running applications. Just a lot of batch jobs to restart after the fact.

[simonlb]

Re: Don't follow the instructions

I've created plenty of documentation over the years and always try to make it so that anyone who has never used it can perform the task being described. That means plenty of screenshots as well as having peer review and having someone who has never done the relevant task to try and do it by following my document before it gets signed off and issued. Yes, it takes a bit more time and there is definitely an art to writing good documentation - anyone ever tried using a Netware 3.12 reference manual? - but long term, it's worth it.

I even once introduced a new process for a number of teams and explicitly stated in the email to the whole department, to read the document and then read it again. One person who always skimmed through instructions was the first to come and say the process was wrong: I told them to go away, re-read the document then retry. Only had three people say there was a problem, but they had all skimmed through the document and missed a step which was highlighted in bold text with a screenshot.

My team leader walked past my desk later the same day, smiled at me and said, "Good email that mate."

[nintendoeats]

Re: Don't follow the instructions

There are many many reasons to write documentation. They are the same reasons that your operations will benefit from having a process and culture that respects the importance of said documentation.

I believe that when people say "documentation is pointless because it's always wrong and nobody reads it", what they mean is "documentation is pointless because I can't be arsed to do it correctly, and by the way that means you shouldn't bother reading any of mine". It's a self-fulfilling prophecy.

[J.G.Harston]

Re: Don't follow the instructions

On most jobs I've been on, I've started with rewriting the incomprehensible and outright lies in the documentation.

[heyrick]

Re: Don't follow the instructions

"outright lies in the documentation"

Ah, so the documentation was written first.

[Vincent Ballard]

Re: Don't follow the instructions

Sometimes the documentation was written second, but the system has changed underneath it. I've got a document which explains how to do a fairly complex sequence of steps in the Azure Portal, and every time I use it I discover that the Portal's UI has changed.

[John Brown (no body)]

Re: Don't follow the instructions

"Azure Portal, and every time I use it I discover that the Portal's UI has changed."

Ditto for pretty much everything MS related since auto-updates have been enforced, but most especially anything cloudy.

[nintendoeats]

Re: Don't follow the instructions

IMO, writing either "first" is a mistake.

If you write the documentation before writing any code, what you are creating is an idealized version of the program in your mind. Reality is going to bite you in the ass, yo often won't be able to implement the API or process that you want/

If you write the code before writing any documentation, then you are probably going to have an API or process driven more by convenience of implementation than the quality of experience you are delivering to the end-user.

I think the best thing is to write them in tandem with one another. Figure out how the feature can be implemented, write some documentation based on that code and decide whether it makes you feel like a war criminal, modify the code to match something better, repeat until both are finished at the same time.

[rcxb]

Re: Don't follow the instructions

Day 1, I followed the instructions, erased all the files on the user disk [...]

My supervisor said (face palm moment) that is an example of the command - not the actual command.

The documentation was changed to make it clearer

You're anecdote reminds me very much of some the users I had to support, years back.

Day 3: You're NOT supposed to type the quotation marks around the command.

Day 4: You missed the step that tells you to exit the editor. The commands don't work because you're just typing them all into a text document...

Day 5: The word "replace" means "put the disk back in," NOT "throw it in the trash and go get a new one."

Day 6: Nothing wrong with your computer. It keeps rebooting because you keep pushing the reset button when you mean to push the floppy disk eject button. Yes, it did work fine yesterday, because yesterday you pushed the correct button... It's not the computer that forgot how to do its job from one day to the next.

[simonlb]

Re: Don't follow the instructions

push the floppy disk eject button

I've just realised, I've never, ever seen a floppy drive with the words 'Eject' near the eject button on any of the PC's I've used over the years which had them. I've seen 5.25" floppy drives with a curved arrow to indicate which way to turn the lever to eject those disks, but never a label on the front of a 3.5" floppy for ejecting the disk. That includes the few hundred I built when working in a shop in the mid-nineties.

[rcxb]

Re: Don't follow the instructions

I've never, ever seen a floppy drive with the words 'Eject' near the eject button

The Japanese Kanji symbol for "eject" *just happens* to be indistinguishable from a rectangle...

[heyrick]

Re: Don't follow the instructions

Day 5: The word "replace" means "put the disk back in," NOT "throw it in the trash and go get a new one."

To be fair, it's a perfectly logical assumption given the meaning of the word "replace" to the average person. It should be better worded to avoid such ambiguity, something like "reinsert the disc that you used previously (in steps 4-7)" or something.

[rcxb]

Re: Don't follow the instructions

better worded to avoid such ambiguity, something like "reinsert the disc that you used previously (in steps 4-7)" or something.

But after I insert the disk by step 4, the manual doesn't close properly anymore...

[Bebu]

Windows 95...

"Absolutely ghastly. Just don't even talk about it."^*

Then being on the Unix side of the loony line I only had a nodding acquaintance with W95 but I think it still used a FAT file system and that chkdsk would be used to detect cross linked clusters.

The FAT is basically a linear linked list of fixed sized clusters (of contiguous sectors) with the index of the list head stored in file's directory entry - like a very poor inode implementation. If the indices in one file's linked list get scrambled with another file's (or directory's) cluster list bad things happen.

Think a unix subdirectory ("/home/simon/bad")^☆ hard linked to "/" and "rm -rf /home/simon".

Precambrian Unix lacking mkdir(2), used mknod(2) to create a directory and using link(2) to enter ".." and "." - both restricted to root. /bin/mkdir was suid root I think. I think as late as SunOS 4.x hard linking directories was still possible with "/usr/etc/link."

^*HHGTTG

^☆If the BOFH were clearing his desk I would be quickly heading to the airport...

[Stuart Castle]

My old college lecturer was doing some design work for a major bank in the late 80s. I forget if she even told us which one, but it was one of the big 4.

She was working with the team that maintained the systems used for credit card processing. They had a major failure, but took regular backups.

The previous day’s backup was unreadable. Luckily the one from the day before was fine.

They lost a day’s credit card transactions. Lovely if you happened to buy anything on your card that day. Not so lovely for them, because it would have cost millions.

[K555]

Those were peak 'fix by reinstall' days

Yes, it's still a bit of a go-to when you're stuck. But 95 dictated you reinstall it so often you could almost benchmark how good you were at day to day PC work by how quickly you could reinstall Windows, load drivers (knowing which ones worked being a massive time saver) and de-junk it ready for the next 3 weeks of operation before you needed to format it again.

In that context, it makes the current iteration of Windows look palatable.

[Mast1]

Re: Those were peak 'fix by reinstall' days

Ah, are you hinting that we have progressed from something like "fish food" releases to "dog food" ?

[FirstTangoInParis]

Re: Those were peak 'fix by reinstall' days

ISTR that by the time we were ready to migrate from W95 to XP, my laptop was crashing twice a day, regular as clockwork. XP was marketed as 13 times more reliable than W95, so I looked forwards to it only crashing once a week or so. And he things W95 allowed us to do as mere users .... 'can you put this CD in your drive and share it on the network so I can boot this thing which doesn't have a CD drive? Yes of course ....'

[Dave314159ggggdffsdds]

Re: Those were peak 'fix by reinstall' days

Ime the vast majority of crashes were caused by shitty drivers rather than Windows itself. Laptops were an utter nightmare back then.

My dad had got in the habit of building home-brew PCs to a budget using lesser-brand components, which we cured him of when he tried a non-budget system using name brand stuff that crashed half as often.

If you could manage with a pc that was just motherboard, processor, ram, established basic graphics card, etc, it was relatively stable. As soon as you added anything niche, or new-tech that hadn't had the bugs worked out, the crashes came back.

Bizarre as it may seem, this is still something of a problem today. Take the space shuttle approach and steer clear of anything new to the market and you get very stable systems. Try the bleeding edge stuff and things don't work as well.

[rcxb]

Re: Those were peak 'fix by reinstall' days

the vast majority of crashes were caused by shitty drivers rather than Windows itself.

Windows needed drivers to run.

Windows did nothing to ensure drivers behaved.

Windows crashed.

OpenVMS ensured drivers behaved.

OpenVMS didn't crash.

[Jou (Mxyzptlk)]

Re: Those were peak 'fix by reinstall' days

> OpenVMS ensured drivers behaved. OpenVMS didn't crash.

Which is quite easy if you shrink the list of "supported hardware", exclude several classes of hardware, focus one a limited set of functions and features. When comparing it to the Windows world with a similar feature-config: I had Hyper-V Clusters which ran > 1250 days without updates and without any problems. And nobody noticed since they just worked, until I checked and "destroyed" the uptime of those cluster nodes. Highest single host Hyper-V was around 850 days, and noone noticed, until I checked. And I bet there are more out there with even higher uptimes, just forgotten that they need to be taken care of too, like a Server 2012 (non R2) Hyper-V cluster with an uptime of possibly ten years.

[rcxb]

Re: Those were peak 'fix by reinstall' days

OpenVMS' stability had nothing to do with the list of supported hardware, and everything to do with being a microkernel operating system that ran even drivers in an unpriv context, monitored them, and restarted them when they misbehaved with the end user none the wiser as to the near crash.

Memory protections have been a hardware feature since Intel 286s at least.

[Dave314159ggggdffsdds]

Re: Those were peak 'fix by reinstall' days

You know, it's possible to have a conversation about Windows without having to be a Stan for some other system that had other strengths and weaknesses...

[nintendoeats]

Re: Those were peak 'fix by reinstall' days

OpenVMS wasn't usually run on machines where every part used was the cheapest thing you could get your hands on.

[A.P. Veening]

Re: Those were peak 'fix by reinstall' days

Try the bleeding edge stuff and things don't work as well.

There is a reason it is called the bleeding edge.

[Dan 55]

Re: Those were peak 'fix by reinstall' days

In that context, it makes the current iteration of Windows look palatable.

It still is peak "fix by reinstall" for those who have their computer stuck in a boot loop or worse by the January 2024 update. Still not withdrawn, still not reissued, official advice is the user should resize partitions, but the instructions are wrong if you're trying to fix it from the pre-boot command line.

So... amateur hour, yet again.

[Anonymous Coward]

I think the problem was originally user profiles went into the C:\WINDOWS\PROFILE directory tree and by default all apps would save data into subdirectories of the user profile dirs in there so DELTREE C:\WINDOWS was probably deleting all the users data as well as the windows install.

Setting up a separate default dir (normally D:\HOME\USERS ...I liked to have this on a separate drive/partition as well) was what I did precisely to avoid this sorry of issue. Subsequently MS have also realised that this is a good idea!

[Lee D]

Pity that ever since they've gummed it up with absolute junk and allowed applications to just dump everything they like into it, when that sort of stuff should be way outside the user's DOCUMENTS folder.

Don't even get me started on "3D Models".

I've taken to creating a "My Actual Documents" folder inside My Documents (or is that Documents, or is it the home folder on my OneDrive synced folder, or is it C:\Users\Username\Documents?) so I know where MY stuff is and can just ignore all the other junk that does not get redirected properly or gets created against my will.

It's like having an office and anyone is allowed to come in and just throw any junk they like on your desk and keep recreating that junk when you remove it, and there's nothing you can do about it.

[Jou (Mxyzptlk)]

Oh so you complain that Microsoft missed the chance with Windows 95 to encourage and enforce "APPDATA" and especially "GAMEDATA" right from the start? Welcome to my club! It took up to Windows 11 until they made the GAMEDATA, aka "Saved Games"... A bit late...

[nintendoeats]

...You actually put stuff in the My Documents folder?...

I've always preferred E:\FilesNStuff (or these days nfs://BombassServer/FilesNStuff). If Windows thinks a folder is special, I definitely don't want my files anywhere near it :p

[Will Godfrey]

I remember Windows 95

It was the first, and last version of Windows that ever sullied one of my computers.

[Jude Bradley]

Ahh, the good old days of working in first-level support for a PC manufacturer who once sold their products in black and white boxes.

[skswales]

Our company had the misfortune to purchase one of their systems. Trying to fix said system on receipt, got through and was put in what I can only describe as a conference call with around six other folk support were trying to help at the same time. Support were so useless some of the callers got their fixes from the other callers.

My issue was hardware related. Monitor was poorly focused. Support's solution was to only run it in VGA mode. It was a 21" monitor.

[Jou (Mxyzptlk)]

> Monitor was poorly focused.

Get a screw driver, preferable isolated handle. All CRT monitors (and most CRT-TVs) have their focus adjustment exposed on the back through a hole. You should check unplugged (and off for a few minutes) to have the feeling to meet that exposed control. Once found you can leave the screwdriver there, and turn the monitor on. And then dial until you have the best possible focus.

[Anonymous Coward]

I think you're about 2 decades late with that advice :). Good advice, but now LCDs come with their own problems (although it's been a while that I've seen one with a defective pixel - that used to be fairly common with the early LCD screens).

[John Brown (no body)]

"All CRT monitors (and most CRT-TVs) have their focus adjustment exposed on the back through a hole."

Asa field tech of the relevant era I can state with a fair level of confidence that CRT monitors of the DOS and/or Windows eras rarely had access to the focus control from the outside of the casing. Early ones from the 8-bit days and maybe early 16 bit days, sometimes did, but not always. Mostly it was internal and part of the LOPT.

[Jou (Mxyzptlk)]

I must admit: I am not sure about the "All CRT" right now. I know SOME have, since I did the adjustment on some without taking the back cover of, but "all" may indeed be a stretch.

This artice, pic three is the adjustment I talk about. Another one here. As for the other things inside the CRTs for focus adjustment: I tried them, but always returned them to their default (marked) state since that was the best result.

[John Brown (no body)]

Yeah, both the links show the device with the casing removed. The first one might have access with the cover on since they made access through the chassis too, but generally, since the focus and HT voltage adjustments are on the LOPT (AKA flyback transformer), it's generally discouraged from being user accessible since they may try to use metal tools and potentially have it be the last thing they ever do :'-)

I do remember an old NCR[*] 12" b/w CRT screen which did have all adjustments accessible from the outside and possible some Philips green screen mono display commonly used on computers in the early days. B/W screen use a much lower HT voltage than colour, and safety was becoming more of an issue, so that's probably why I don't recall ever seeing a colour CRT with user accessible focus and "master brightness" (ie HT voltage) adjustment.

[*] Yes, that NCR, the cash register people.

[AustinTX]

Cleanest install evar

It used to be routine (though bad practice) to run program installers from the root directory. Software would typically unpack into a temp directory, install the software, then clean itself up. Which is what I was used to and relied upon happening until the day I installed a major software package which cleaned up the lazy way by just deleting everything in the directory it was run in. I called those fruckers up and chewed them out.

[Grogan]

What? Windows 95 would just reinstall over top of what was there by default. It wasn't so complex that it didn't work, either (might have to install any patches if you had... like the winsock2 update etc.) and of course if the computer was infected with a virus (they were actual file and boot record infecting "viruses" back in those days) you wanted to do more than that. More than just format in the case of a MBR infector too.

I couldn't count the number of times I reinstalled Windows 95 over itself on PCs. Yes, including fixing cross-linked files on FAT with scandisk first then reinstalling to fix corrupted files. That even happened to me personally once. What was I doing... testing filling up memory and the swap file until it crashed and it somehow corrupted my filesystem lol

That became less practical (and less successful) with Windows XP (NT based). Often if Windows XP was buggered, a "repair install" or even an "in-place upgrade" to reinstall it would fail.

If you deltree (or install Win9x to an alternate directory) applications were disconnected anyway. You'd have your files, but stuff like Microsoft Word wouldn't work and would need reinstalling. Better to just clean that up too. (and there will be crap in c:\program files\common files). Some applications would need reinstalling even with a successful reinstall of Windows 95 if they updated dlls that got overwritten back to older versions. That was a gotcha back then, applications often overwrote files in c:\windows\system and friends.

AND, "tech support" shouldn't be coaching end users to do things like this over the phone. Come on... I knew technicians that were scared to use deltree themselves. Working on a damaged filesystem with anything can cause corruption though.

[SteveK]

I once raised a support call with Dell on a machine with a hard disk that was starting to misbehave and report errors, probably in the W98 era. The tech wanted me to wipe and reinstall, which I was reluctant to do, not for fear of losing data which was all backed up, but just because it was clearly starting to fail and it would be a waste of time. He then sent me some instructions to do some diagnostics, which involved running debug.exe and entering a bunch of hex and running it - it seemed a bit excessive, but went with it - whereupon the 'diagnostics' nuked everything.

Subsequently deciphered the hex back into x86 code to find it simply stuck some numbers into registers and jumped to some BIOS system call to overwrite the boot sector and partition table with zeroes. As expected, the reinstall didn't magically fix the faulty disk that was clonking away to itself and which was duly swapped, after unnecessarily wasting several hours of my time. Not happy.

[Grogan]

I've heard of tech support coaching people to use the low level debug DOS utility. Not just erasing structures on disk, but clearing CMOS too. I would have liked to reach through the wires and do three stooges stuff to them. You have to know exactly what you're poking at when writing to registers. You can corrupt firmware if you knock on the wrong doors. Back then, "flash bioses" were only in use for a few years. It was possible to corrupt it... a virus could do it by writing one byte to the bootstrap code in the bios to make the computer unable to boot from anything. You could brick a computer with debug.exe

P.S. There was a time when you could corrupt EEPROM on certain IBM Thinkpad laptops, simply by probing for sensors. The developers wrote in a "Thinkpad detection" routine after that, where the application (lmsensors' "sensors-detect" program) would quit with an error if any Thinkpad was detected :-)

[AustinTX]

Speaking as a former Dell CST... whatever the Dell tech told you to do, it was to get you off the phone because the call was ruining their average call times (ACT), and there was NOTHING higher priority to their manager than short calls. NOTHING. When I worked there, I was ordered to take my next call without logging any details about the one I had ended. I was told to "go back and put in the details when you have free time". As if one could remember that degree of detail hours later. There was never any free time. You can imagine how many customers called back with results from the pointless busywork they had been assigned only to be told there was no information in their call history and they would have to go through mandatory dumbass troubleshooting yet again.

[biddibiddibiddibiddi]

I would never have considered running chkdsk in order to specifically verify that crosslinked chains didn't exist in order to ensure that deleting a file wasn't going to accidentally delete other files. Running chkdsk with a FAT filesystem in particular is just a good idea in general when you're having problems, should have been the first step, and likely would have fixed the boot issues in the first place. This was outright negligence and stupidity.

[Dwarf]

I had something similar happen

Back in the very dim and distant past, whilst using Unix - on an AT&T system with a 68000 cpu and Xenix on 286’s, I got to like the rm -r command and was frustrated because DOS didn’t have an equivalent, so I wrote my own, it was called prune as you could prune branches of the file system.

During development on my shiny new 486DX2-66 with Borland C, I was getting my head around recursive code that could wander around the file system, I learned a very important lesson of handling the . and .. folders. This was after a test on a small folder, where it was taking a lot longer than expected, it had wandered up to the top of the disc and wiped virtually everything. Luckily, my new machine also had a QIC tape drive in it and I had backups, so there was plenty of time to re learn about .. whilst the drive was rebuilt from tape.

[Captain_Cretin]

Hmm, I smell a rat.

Win95 was my first exposure to WinOS's and I think I reinstalled it about 50 times in the 1st month, paring it down to under 40 minutes, as I messed around and broke it.

It was quite possible to write a fresh install over the top of the existing install and lose nothing; as long as you hadnt stored anything in any default "My" folders.

To this day, I never use those folders !!

I forget the exact process, I had a printed sheet of instructions from Evesham Micros that had arrived with my mum's new PC - which was the reason I built my first Win PC - to sort out WTF she was doing to hers (saving everything to the root of C: was the main cause, it turns out).

[Stuart Castle]

When I used to work in a university computing lab, the bulk of the pcs in the lab ran NT4. Apart from one. This, because it was required to control a scanner that only supported

Windows 98, ran Windows 98.

On average, a new install lasted about a week before being unusable.

Not surprisingly, the second HP released NT4 compatible drivers, I upgraded the machine to NT4.

[Jou (Mxyzptlk)]

Yeah, you cannot protect Windows 98 from malicious users. With NT 4.0 is, in comparison, very easy to secure.

[RAMChYLD]

Ah yes. Winblow$ 95

The beast that needed to be reformatted every 6 months if you mess with a lot of shareware and stuff unless you are patient enough to wait 15 minutes for it to boot, and if a bad driver update didn't do it in first. Back in those days the common hard drive (Quantum Bigfoot) spun at 2600rpm and operated at IDE PIO-2 mode so you'll definitely be watching data crawl into the system.

[nintendoeats]

Re: Ah yes. Winblow$ 95

The Quantum Bigfoot was also known for being intolerably slow even then. There were much faster drives out there.