The memory is hazy...
...but under certain circumstances, didn't Windows 98 used to come up with this example of error message brilliance from time to time?
When I was young I built up a collection of system error messages. Ok, look, it's not as sad as collecting stamps! It is? Really? Oh well, never mind. Anyway, my recent piece about Borland putting rude words in Quattro Pro got me thinking it was time to revisit that collection. Some of them date back to the days of the mainframe …
...but under certain circumstances, didn't Windows 98 used to come up with this example of error message brilliance from time to time?
I learned my favourite error message when I first started messing around with linux (about 12 years ago now, I guess.)
I was trying to use my linux box to act as a networked printer server for my parallel port epson. However, the first time it ran out of ink, the lpr daemon reported:
"Unknown error. (Printer on fire?)"
You should have seen the speed with which I ran across the room to ensure this wasn't the case.
But not exactly creative. My favourite still remains Commodore's "Guru Meditation Error". It was very zen.
Tux, because the other day I actually was able to read a kernel panic and understand why it was panicking.
My favourite from Multics Emacs... http://gizmonaut.net/blog/software/multics_source_reveals_le_jetteur_des_gazongues.html
So very fortunately, those days of constricted memory and disk storage are far behind us. I have not see messages like these:
"Failed to open file" (or any of numerous equally uninformative variants)
"The parameter is incorrect"
"Your socks don't match"
since, oh, last Saturday, though in all honesty, that may be because I wasn't doing anything here yesterday.
The handler was called "Dire Straights" because it had the same initials as the "original meaning". It was used in the 68k versions of their operating system (two CPUs ago). I don't know if it has been carried over, but...
On another front, the ROM monitor of a machine I have (before I changed the ROM image) had the error message "WTF?" which was emitted when it couldn't understand the console entry. The official explanation was "With Trace Flag" (and I have some swamp land to sell!).
IBMs error messages from the OS/360 era usually had a nice 6 character message number (prefix indicated which program, the other stuff told what). They had an entire book of error messages. It was before the idea of "internationalization" took over (it was the 60's).
My brother did get the classic "Drive C is full. Please Insert another Disk and Try again"
With only an OK button.
And "blem wit" of course is matched by the completely empty windows dialogue box with just an exclamation mark in it and an OK button
My wife is a (now-retired) banker. The mainframe system at the little bank where she worked in Texas had very, very good input verification, but once in a while someone would enter something from which it simply couldn't make any sense. When that happened, the message returned was:
Input respectfully ignored.
For terseness nothing beats the TRS-80 error messages, which consisted of: "What?", "How?" and "Sorry".
No discussion of cryptic error messages is complete without at least mentioning ed - the forerunner to the Unix vi editor.
Any error on iput generated the one true universal and cryptic error message namely
Yeah, I admit it, I'm the kind of guy who runs strings(1) on an executable file. So when I ran strings on FrameMaker.exe way back in the mid-'80's on Sun's BSD Unix, mixed in with the other error messages was my very favorite one. I don't know what extremely desparate and unrecoverable situation evoked this message, but the text was
"Evacuate now, in our moment of triumph? I think you overestimate their chances."
Remember that old joke: "If Ken Thompson designed a car it would have just one light on the dashboard which would light when there was a problem. The experienced motorist would know what the problem was". Yes the early UNIXen had to squeeze down into an early PDP-11's memory space (64k code and 64k data?).
...from Excel VBA:
A colleague of mine was at pains to use a Windows box to build some software we were working on. We had to test audio playback but the card in the machine hadn't been installed properly. So he installed the audio drivers and restarted Windows, and on boot-up, there was a problem and the driver could only play the first tiny fragment of the Windows startup sound, which was cut short and followed by an error message dialog saying the sound driver couldn't play the sound.
The funny bit was that this error message dialog was accompanied by the Windows alert sound (alert.wav), which was similarly truncated, and resulted in a second error message dialog, triggering another truncated alert.wav, and another error message dialog, etc... etc...
Most amused by this, we left the poor Windows box filling the screen with error dialog boxes and went for a long lunch.
Task not in system
as favourite error message. Given th e"Local" (to your TERMINAL) status of every file you edited, which you then first had to save AND make permanent, any jitter on the line could easily wipe out an hours work EVEN if you saved stuff. This frequently caused th efollowing dialogue
$ F**K YOU
Task not in system
In these chilly mornings my wife (a usability designer) is amused by the screen on her car radio which pops up a picture of a snowflake and the text "3°C Warning Icy Road" with only an "OK" button. It certainly isn't OK, but what can she do?
I'm surprised no-one's mentioned
"No keyboard found. Press F1 to continue" yet...
..was after I broke (for the 10th time) my xorg install on Linux. It crashed, and before the compute rebooted and after a bunch of kernel stuff it said "Serious error. Uhh, your system is seriously misconfigured."
It was then I decided to reinstall..
this one can be found Windows v1
Press any key to continue or any other key to quit.
Solaris has my favourite error message for cat:
$ cat flap
error: cat can't open flap
A certain genealogy program I use has the opposite of a terse error message!
"Find in Diagram" does nothing when you are viewing a diagram. To use it, switch to another window, but without actually closing the Diagram Window. Select a record in the other window and click on "Find in Diagram". You will be switched back to the Diagram Window, and the selected record will be located (if it exists in the diagram).
Pheeeeewooh! Oh, and there is only an OK button, and when you click the OK button all those nice instructions disappear so you better have a good memory!
But "Keyboard Bad, Press F1 to Continue" is still a classic.
For me the best was the Amiga
"Guru Mediation Error"
I always pictured an overweight monk deep in mediation suddenly farting.
My personal favourite, produced by a compiler for a language that allowed type definition:
"Integer expected, but rabbit found."
I think it was when you tried to login to the "Nobody" system account like it was a normal user account: "You don't exist. Go away."
Still see on the rare occasions out of Tivoli products (with no return or error codes)...
"Something has gone wrong"
"An error has occurred"
Always useful when trying to see what's happened.
Over the course of time, I came across two separate Event Log entries whose description consisted of the explanation:
Error formatting error message.
Never did find out what the original errors had to do with.
On apollo Workstations when dealing with the tape drive :
- unit will not fit thru 25" hatch
they dried selling these drives to the navy for use on submarines. Contract was turned down because they were too big to go throught the entrance hatch of the submarine. They would have had to cut a hole in the hull ( might as well have installed that screendoor in the mean time ... ). Someone coded this as a permanent error in the tapemanager ...
My other favorite thing is an HP54645D oscilloscope ( i am using on right now ) Hit Print/Utility. the bottom menu ( the softkey functions ) will show sofkeys 2,3 and 4 as blank.
Hit 3 &4 to see a nice vector drawing of a badger ( no it's NOT a skunk !!! ) and a namelist of the coders of the scopes firmware.
and if you hit softkey 2 and 3 you can play a kind of space wars .
If you have installed the latest firmware to be able to use extension packs these features are lost. You get the nice message : Sorry There was not enough space in the ROM.
Another cool one is a TEK 2245 oscilloscope. if you browse the built in help and actually read the entry on how to use the help on the functions controls, at the very last page there is this line :
Go ahead Push a button, make my day ...
go to www.eeggs.com for tons more of these hidden and funny items
My personal Top 3 error messages would have to be (in no particular order):
"Error contacting Real Networks Technical Support. Please contact Real Networks Technical Support for more information." (or something to that effect) when using RealPlayer (before it became RealOne, I believe) without an internet connection.
"Amnesia Error: Out of Memory" Self explanatory error from Jazz Jackrabbit 1 & 2 by Epic (then Epic MegaGames)
And the fully fictional error from the Halo 2 video game:
A total FU exception has occured at your location. All system functionality will be terminated.
Press any key to power cycle the system. If system does not restart; scream at top of lungs and pound on keyboard.
If you need to talk to a programmer press any other key.
Press any key to continue."
There is a point in the game where you must use a computer console to open a door. Continuing to use the console after the door has been opened results in the screen turning blue and displaying the message above.
My Dell Inspiron D810 running Windows XP SP2 produces the same error message: When running on the battery and the battery power level is low, the computer is supposed to produce a warning message. Instead, it immediately hibernates (presumably because the battery is poked). When I restore mains power, you can see the no-text message sitting in front of the login window.
I recall that in the early versions (6.0?) of the Mac OS that ran on laptops, there was a dialog that was supposed to display when battery power ran low. The text for the message was held in a resource, which often wasn't in memory when the message had to be displayed. So in order to display the message, the hard drive would have to spin up, which would completely drain the battery of any remaining juice...
Something similar I guess.
Paris Hilton icon, because I haven't used her yet, which probably puts me in a small minority...
Although not strictly speaking an error message, I loved the information message issued by IBM's mainframe error-reporting program EREP:
"History in progress..."
...had the delta debugger. With this utility you could patch the contents of RAM of a running system - by specifying hex addresses and values. Given the potential for disastrous consequences, this was not a tool for the novice. Perhaps for this reason it had no command prompt and only one error message - "eh?"
if those errors had been in a TSR, could they be described as a trs TRS TSR err?
Many years ago, sometime in the last millennium, we had a Burroughs patient result system which we used for research. The search language was called TCP (Test Control Protocol) and it was made up of very primitive commands (Move pointer, compare result, advance record, that sort of thing. Stuff you'd do now with SELECT * FROM MyTable WHERE Surname LIKE "Smit%" took hours to code!)
Every move or search command required two destinations - it jumped to the first if the search/move succeeded, and the other if it failed. Just about the first instruction in every search program "Go to start of database" - I can't remember the syntax, but we knew the ref number of the oldest specimen in the database, and coded it in. Now, that jump couldn't fail, but the syntax demanded a fail destination, so I coded one in.
About eight years went by, and the database grew (we worked out later) to a size which meant the compiler needed to be told to use a bigger data type to handle addresses - but we didn't realise that until one day when the computer couldn't make the jump to lightspeed - sorry, to the first record. And a rather confused lab tech brought me the sheet from the lineprinter: UNKNOWN AND UNFORSEEN ERROR. SPOOKY, ISN'T IT?
When the C: drive is full, attempting to delete stuff gives the brilliant error:
Error: Insufficient space to delete files, delete some files and try again.
There are lots of good error messages around - TeX has some doozies - but nothing as elegant as the haiku ones in NetPositive, the BeOS browser. Howaboutthis, for server timeout:
Server's poor response
Not quick enough for browser.
Timed out, plum blossom.
For me was a terminal app I used to use on Win'95 back in its early days.
Occasionally, it would just terminate the connection, and lock solid until you pressed the "OK" button (the only button) on the dialog box that read "Error: No error has been detected.".
It would then terminate.
...but we once managed to ship some code that had a message box "Haven't written this bit yet"
From a previous existance and left for our entertainment by a programmer who had long since left the company concerned.
Green-screen app with what can only be described as a shed-load of input capable fields on the particular screen in question. All are validated against reference data and each other and any in error are highlighted and some nice scrolly messages appear at the bottom of the screen telling you about all the errors.
One day a new user calls saying that her screen is locked out, we ask what she was using and swear (offline) when she tells us. The particular screen in question had a few sensitive things open for update. So, fact-finding time:
"What's the screen number?"
"It hasn't got one."
"No, look in the top left corner of the screen, there's a screen number."
"No there isn't."
"What does it say on the screen?"
".......I'd rather not say."
Hmmmmm, very strange. A quick look at the user's job reveals a screen display member that we don't immediately recognise. Do we a) pull the source to see what it looks like or b) exit, en masse, to the user's desk to see what's up? b) wins out and off we go.
Turn out that the new user has managed to get every single field invalid in some way or another and a long-unused error routine has creaked into life. There, on the screen, with all the keys disabled, resplendant in HUGE block letters lovingly constructed out of individual spaces displayed inverse, highlighted and blinking is one word:
My favourite was not in English, but in Dutch. When some DOS versions of Norton Commander got given a floppy they couldn't read, they said something like "Floppy in drive a: kan niet geformatteerd worden", /Anglico/ "Floppy in drive a: can not be formatted". This resulted in quite a bit of concern when we tried to read a floppy containing important data in a certain computer with a wonky a: drive, and had certainly not asked for it to be formatted in the first place... until it hit me that some bozo translator with no experience with the program and an unannotated list of messages had chosen the wrong interpretation of "Floppy in drive a: _may_ not be formatted".
I remember in the early days of Demon Internet, when we were all using DOS and slow dialup connections, someone posted "I've just had the error 'Drive C: out of paper', what do I do?" There were several responses saying "Put paper in drive C: of course".
The actual reason was that DOS had a common error handler, and one bit was used to indicate whether the fault was in a drive or a peripheral. Sometimes this got confused, so an actual sector read error on a drive would trigger the equivalent peripheral error message instead.
(I actually wrote an application which said "Press any key to continue or ESC to exit". It made sense to me at the time, my cow-orkers found it amusing...)
My favourite error message which never existed, also from MSDOS days, was "ERROR CLOSING LEVI.ZIP: REMOVE FLOPPY AND RETRY".
My favourite on Win 2000:
Error deleting file or folder - there is not enough free disk space.
Delete one or more files to free disk space, and then try again.
I suspect the reason that the messages such as "Cannot find DLL" are not more informative is that the string comes from a localized resource file. I'm working on a framework at the moment, and I'm trying to put in variable substitution markers in the string resources where appropriate, so the exception messages can have useful data included, but it's a PITA and I can see why people don't bother.
While not as advanced as many others, my favourite error still had an air of accomplishment and being special about it.
Back in the days when creating Doom mods using DeHackEd, if you really messed things up, the game loader would halt half-way through and output
"Professional error. Game halted."
Iow: you messed up, but at least you did it like a pro ;D
Many moons ago I was working on a now happily defunct public sector financial management app when we we began to get reports of the following (fatal) error from windows 95/98. "Unknown error 5000001".
If it's unknown, how the fsck does it have a number ? What are the other 5000000 'unknown' errors ???
This was of course our brutal introduction to the hideous painful world known as DLL Hell.
From the other side, I often use fairly humorous error messages in code I've worked on, many times including the word "b0rk". If the user is looking at an error message, the least you can do is try to stimulate their sense of humour before they pick up the phone.
I did once receive an error message from a Windows server along the lines of "The server is unwilling to service your request at this time".
I was torn between fear for the future of humanity as the machines rose to power and hope because the server was not ruling out that it would be willing to help me at some future time.
A colleague of mine got in serious trouble when one of his Pratchett-esque error messages flashed up on a customer screen:
"Out of cheese error"
Though there is at least one system out there that suggests that people "Eat More Lard".
My first attempts to really muck around with Linux came to grief with the following message which required a reinstall to get past:
"According to /var/run/gdm.pid, gdm was already running (616), but seems to have been murdered mysteriously."
We used to use an accounts package called Lakeview, a reasonably small company but written and supported totally in house.
We saw a few classics, such as:
"Lakeview is confused"
"An error that shouldn't occur has occurred"
Sadly we're on to a new system now so i don't get to see them anymore :(