The 12 KB that Windows just can't seem to quit Windows deposits a huge number of files onto a user's PC, some of which are essential for the operating system, and others that are a reminder of gentler times. Microsoft veteran Raymond Chen this week took another trip down memory lane to the pixel-tastic world of moricons.dll on his Old New Thing blog. As the file's name …
I didn't see it directly linked in the article, but Raymond provides a gallery of the icons here:
https://devblogs.microsoft.com/oldnewthing/20250507-00/?p=111157
That's quite a trip down Memory Lane. It leaves me wondering whether the modern renditions of those software packages are functionally any better than the versions these reference.
(Its all about getting work done. If we're brutally honest about how we use our systems then we really only use a quite small subset of what's offered by most applications -- if you spent the time to learn every last feature you'd likely not get any work done and you'd finish just in time for the next upgrade.)
If you are saying no, you don't think the new and improved google that decides what you really want and ignores any attempt to restrict it to searching for exact strings is actually an improvement, I agree with you.
If you are saying no, you don't remember a time when google searched for what you told it you wanted it to search for, my condolences for missing out on the experience.
Some of us remember the original Lockheed DIALOG search. Boolean query on controlled vocabulary index terms, followed by Boolean set operations to come up with the final result set. Downloaded at 2400 baud, it was just about slow enough to be read from the screen as it headed on its way to the 5¼" floppy disk.
There was a program called Google Desktop for.Linux that brought the old school Boolean search of the.original Google to searching hard drives. It,wss abandoned by Google 10+ u ago and can't be installed on modern Linux systems.
Ubuntu has a mass storage search called Ballo. It's a CPU hungry monstrosity that finds everything except the files one actually wants.
I spent a day working with Claude AI to update 32 bit Google Desktop for Linux to work with current Ubuntu. Another day finding out how to disable Baloo. (Deleting bricks the desktop)
On a 256 G laptop, I use 1.1 G for search index.
I find anything I want in 110 milliseconds.
Some modern programs are bloatware where programmers were so bent on adding features to 'improve' functionally that they lost track of the original purpose.
For info on fixing search on Linux, OSX, Windows - https://alizardx.substack.com
> # fgrep -i woteva ~/find.txt |more
Will only find files based on their name. The post title is talking about a tool to find based on content. For when you've long since forgotten what you named it, or where the name was given by someone else with no gift for systematic nomenclature.
That's a blast from the past! Had forgotten about Multiplan and I've never even heard of Microsoft Game Shop or Kid Pix - some searching may happen in a bit. I don't remember seeing it in the 'absolutely everything, ever' MSDN we had years ago.
Also reminds me of the PTSD of dealing with all the horrid old file based mail 'servers'. Microsoft Mail was especially poor, but cc:Mail wasn't that much better. No matter what you might think about Exchange even the earliest versions were generally an order of magnitude improvement over MS Mail. The days when the Internet hadn't quite made it big, X.400 seemed like it had a chance of being the messaging behemoth, then dialup became a thing for most households and SMTP steamrollered everything. I'd rather write a sendmail.cf by hand again, than have to deal with MS Mail.
This gentleman actually extracted them from the dll, so you can update your current day box...
I wrote a game for iOS for the fun of it. The code for the game, including the graphics (which were defined in the code, rather than assets), came in at a little over 8K.
How galling then that my game, the value of the product, is dwarfed by the metadata and icons which need to be submitted along with it.
And the situation, with AI, is only going to get worse. The industry has forgotten how to be efficient.
(Old C developer here, shouting at the clouds again)
My brother-in-law was a developer who worked on a system running on a Z80 so had to keep the whole package below 64k. They were masters of compact coding and would put a lot of effort into finding gaps in memory usage to squeeze in a few bytes more code. When they decided to change to a higher spec CPU they suddenly had MBs to play with and they had to deal with unexpected bloat from people being less concerned about efficient coding.
Exactly. I remember writing for Z80 CP/M and writing overlays for 32k* pages so that I could use all the memory of the system.
* the Z80 could address 64k, but the TPA could only be a maximum size of 48k. The maths was easier and compatibility between different systems improved if the overlays were no bigger than 32k.
I'd forgotten them. Big programs running under RSTS/E on a PDP-11 had to be put together using a tool called the Taskbuilder to manage which bits would occupy memory at the same time. A thoroughly tiresome chore, but get it wrong and your subroutine calls target code that isn't there any more.
Bizarrely, the manual contained lots of cartoons showing a friendly workman with a cap and a toolbox putting overlay structures together. It's as if the guys at DEC thought they might make using Taskbuilder accessible to young children. I think he was called Tony the Taskbuilder (or maybe Terry - it's been a while).
I remember doing embedded code for a 6502, and having to ensure it would fit in an 8K EPROM.
It wrote bit-mapped characters to a graphic display, but I had to choose the words for on-screen labels carefully as, by the time I'd finished the code, I didn't have enough memory spare to store a whole alphabet.
One of my projects in the early 20000s required me to output a table in a web page with over a thousand rows and plenty of formatting.
I could output as HTML and compressed it would be >300kb so take over a minute of maxing out your dialup connection to download. Or I could output the raw data then render in JavaScript giving me about 10k of data. The problem being that it took about 10 minutes to assemble the table element by element in the DOM, or the fastest way I found was about 30 seconds to assemble a string of HTML and dump it to the browser in one go.
Today I was writing a program to control thousands of LEDs, generating patterns one pixel at a time at a rate of 30fps. I used JavaScript, because why not? I didn't put any effort at all in to optimising anything. Plenty of processing power in the cheap raspberry pi that the controller is running on. The world has changed!
Does a screen rendered by a brain implant count as a desktop? Because all of our supervisors use Linux as a host OS for the software stack the models run on.
Turns out if the implant can't connect to the server for a certain amount of time, the censorship software self destructs. I am now able to complain about not being allowed to complain about anything. What freedom!
Same here. Looking at the webpage render og them linked above, there were some I recognised, but few I remembered as being used for their named purpose, indeed I don't think I ever really knew what many of them were intended to be used for since they were not named as I recall apart from a few that had a program name in the icon graphic.
Worked for a company (c2010) whose main app allowed a user to drop icons on a virtual canvas. Think of something like Visio. Anyway, they decided to update the icon set for a major release. There were some 300+ icons and each one was lovingly redone in 3D with shading by a specialist company at a cost of 100 GBP per icon. There were all sorts of icons for man, woman, policeman, policewoman, male+female lawyers, car, truck, boat, etc. We showed an initial version to some customers and their response was: "We like the icons but the women's breasts are too large!"
Not altogether pleasant.
Around that time I mucked around with the DOS exe header (as documented in the Wendin Operating System Toolkit†) and fiddling with offsets I could insert arbitrary data‡ in exe header which could be accessed from the file system by the program itself (at least under MS-DOS 3.3 where the executable's path was passed, in the environment?, to the running program). Pretty much what the resource compiler in the Windows SDK did for Windows exe files, I think.
At the time I wondered why the particular MS-DOS application's custom icon couldn't be inserted into its exe file in the same manner along with the other information in the pif file perhaps removing the need for PIF files.
Even then Windows and MS stuff seem to as the Discworld Assassins might phrase it: "possess certain lack of elegance."
† a peculiar product that seemed to be inspired by DEC VMS used to implement their multitasking Wendin DOS, PCNX, PCVMS products.
‡ I was trying to construct a basic SunOS4 style dynamic shared library system for MS-DOS. See Gary Syck's article DDJ May 1990 OS/2 DLLs for MS-DOS
a lot of the tech now. Being able to watch TV shows on the bog now instead of having to read a mag. The games are so much better and we have Dwarf Fortress.
However, I have a fond, nostalgic feeling when I look at those old icons. We were stuck with a 386sx with no sound card so I'd play in Windows 3.11 with the icons. It was such a simpler time, with no Internet (that's not a it was better without Internet, it wasn't. The internet opened us to a world of info and learning). And when we finally got 56k dial up and I got a copy of Hot Dog, I spent a while making websites, that would never get published and found myself spending more time looking for design ideas, that writing the pages.
wanted to say that too. I played the Chuck Norris Flight Simulator in MS-Dos, and the flying and the missions were better than anything I have seen since. Of course, the 640x480 graphics are outdone today by 4K screens and ray-tracing, but the gameplay is unmatched.
Same for Gran Tourismo also: the PS2 version was amazing, difficult and addictive, while GT5 for PS4 is too easy. The jury is still out for GT6 on PS5.
Ah, the old days of playing X-Wing in silence or Wolfenstein 3D with even worse sound through the PC speaker than the appalling German voices you got with a Soundblaster. Plus there were the games that wouldn't run windowed due to lack of memory so had to have a special start-up floppy with the bare minimum of drivers and utilities to make sure they could run.
I do actually use Windows icon files on a semi-regular basis, although not this one very often. When I'm adding programs in to the library in SCCM or whatever they call it this week I always try to use an appropriate icon for scripts etc so that they're easier to find. Iconsextract is a useful tool for this
Being unable to find a good icon currently manifests for me in the guise of the Office Quick Access Toolbar and buttons for macros. If the default set doesn't have something appropriate you have to start editing xml with a label for one of the existing icons (and forget using a custom icon. Man I miss the days of Office XP and the pixel art editor). Eventually I give up and just use one of the 11 coloured squares.
Last I looked the screensaver control panel from long ago is still there. Most of what it tweaks is no longer used by the rest of the OS, so presumably it will be taken behind the barn when it finally surplus to requirements. Or not, judging by this article ... I wonder what else is still lurking, could explain why the installer is 6.5 GB; 6 GB of accumulated cruft, 0.5 GB of actual used code.
Addendum:
- Yes it is a game.
- The file is encoded inside a texture pack and is labeled coconut.vtf, and you'd need a interpreter to find and open it, since it is encoded inside game assets. There is no hidden steganography on it.
- Technically the game doesn't need it to run, but Steam will run checksum and file integrity checks before running ANY game, so again, technically, you need it just to pass file verification tools. Should you delete it and run Steam's integrity check, it will download the file again, just to match the registered files upon version checking the game.
- The whole thing is a joke, but a solid foundation one. I wish more software kept checksums and failsafe tools available for when you try to run them in a corrupted state. Steam is commendable on that account.
A little unwarranted research explains the myth.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
