What software do you dream about, but do not have time to code yourself?

One of the tools I have liked but well past my abilities was a image based pdf or scanned document conversion to something tex like as a document or to pdf, such that the output appears the same as the original with the exception any scanned document artifacts or noise is removed, ie resultant output the typesetting and font matches with any images the same or if not sharpened to a suitable clarity. I think it's one of the very few things were a LLM might suit the task well bar the privacy issue. I had a large collection of photocopies I'd done in the 80s, in the order of 1000s (I have somewhere a large box full of them, but moving every so often lost track of them,) and last time I looked over them, scanning to image was just going to take up too much room on a hard drive, I'd rather have had them all on a cd / dvd or two.

Edit - Additional: The scanned page is of course corrected for proper alignment and positioned appropriately.

I'm thinking about two things, or two catalogs of softwares:

1. A 64-bit Windows 2000 continuation project that retains the UI but applies every security patches since then. Is it even possible? Would some security patches need a new kernel? I'm not technically good enough to answer.

(Or, a completely debloated Windows 10 -- by saying debloated I meant a complete removal of telemtries, ads, shops, whatever, and I do not know why I need .Net at all if I don't need the applications)

2. Modern code IDEs based on the same idea, maybe a better Visual Studio but written in native C++, just like a continuation of Visual Studio 6 without getting into .Net.

P.S. my view of .Net is uneducated and biased. I think they slow down the OS and Apps but I have zero clue, so I'm probably wrong. Maybe the .Net code in Visual Studio is as efficient as the C++ code but it's just the product is not modular enough so we have those 10 seconds start up and other slowdown issues.

Some open-source, client-only software which would let me easily save, search, and review all the gibblegabble I write on places like HN or Reddit etc. It would need to capture more than just my posts, since replies need context and that context might also contain the necessary search-terms.

A fully functional Memex[1,2], for the Web.

"instruments are at hand which, if properly developed, will give man access to and command over the inherited knowledge of the ages. The perfection of these pacific instruments should be the first objective of our scientists as they emerge from their war work."

It's been about 80 years... and we still haven't done it yet. This is worse than the debacle with string theory.

A fully functional Memex would work like a normal web browser, but allow you to look back on your entire history locally, and make a copy of any of it for someone else, locally.

[1] https://www.theatlantic.com/magazine/archive/1945/07/as-we-m...

[1a] https://web.archive.org/web/20150214162829/http://www.theatl...

[2] https://en.wikipedia.org/wiki/As_We_May_Think

A simple, idiot-proof app to install Linux from Windows, aimed at newbies. So when people complain that Linux is too hard to install, I can just point them to the .exe.

The app will be a simple installation wizard, maybe using the Nullsoft system as it's simple and lightweight. It'll ask them some basic questions, like what sort of UI/UX they prefer, whether they play games etc, and how much space they wish to allocate to Linux etc. It will then download the most appropriate distro, make the partitions, AND install the distro right from Windows (maybe using WinBtrfs and extracting the rootfs from the ISO onto the partiton), extract and set up the bootloader etc.

The main question mark for me is how to deal with Secure Boot. Maybe I can have a loop at the beginning which waits until it is disabled by the user (maybe detect the system's make/model/BIOS and display the appropriate instructions)? But I also wonder if asking users to disable Secure Boot is even a good thing, it might give them the false impression that Linux isn't secure, or my app is somehow compromising their PC's security. Probably not a good first impression. If anyone has any ideas on how to handle this bit in a sensible manner, I'm all ears.

I wish someone would implement an interface for using color diffusion curves in a vector drawing program: https://wiki.inkscape.org/wiki/index.php/Diffusion_Curves

When they were first proposed back in 2008 they made a big splash, but afaik they never got past the prototype phase. There was even a push to add them to the SVG spec.

There is a piece of software called CloneSpy that is a utility for finding duplicate files. I like the user interface.

https://clonespy.com/features/

I find it useful for finding that I have four copies of a 50MB pdf in different folders, that kind of thing. If I want to free a few gigabytes of diskspace without too much mental effort then a scan to see what the biggest files that I have more than one copy of can work for me.

It is Windows only and closed source.

I presume it calculates a hash for every file then compares the hashes.

I would like a version that I can run on Linux and that nicely handles things like filenames with unusual characters.

A simple tool to take SOX and wrap it for android so I can do all those patches to audio. I can find half the functionality, but nobody wants to give me the commandline audacity "talk like a duck" or "phaser effect done badly" SOX gives me.

Hugin for Android, I think Bimostitch is, but doesn't say so. It only has 3 of the projection models and I can't hand-alter the image alignment.

Another android app: snip this bit and that bit of an image, and join em. Or, snip out a region and shift-up (shift-left, shift-right, shift-down) -Its easy to black out regions, its hard to actually cut paste regions in a photo. Can be useful. Or screenshots.

De-dupe images. Perceptual hash logic. Not EXIF meta data only, a decent model which finds trimmed, edited, EXIF non-compliant dupes in the image set.

An open source, multiplatform disc drive emulator capable of supporting full RAW disc dumps. Subchannel data, error data, anything. Kryoflux-like full sectors dump.

Possibly accompanied by an hw (likely an ESP32 SoC) optical disc emulator, in order to replace any aging drive inside music samplers, multimedia devices or game consoles. (Yes, I know there are multiple ad-hoc devices available, but i'd love a shared developed solution so to avoid having to reinvent the wheel for every single device).

Also, an FPGA core capable of reproducing a 2003-circa x86 gaming pc: a ~500mhz P3 equivalent CPU, a voodoo 3 or equivalent core plus an nvidia tnt2-like GPU and a Soundblaster Live Soundcard (again, this is also inspired by having similar configs aging and not working anymore).

I'd probably need 3 or 4 lifetimes to achieve anything like this, though.

1 - An AI backed captcha replacement that finally stops only bots instead of humans.

2 - An overall OS-level machine learning/AI service, that monitors the user's behavior, and adapts the overall OS behavior to it automatically, and continuously. Kind of IFTTT, but self adjusting, and smart. E.g. if you don't use a button ever, hide it. If you use an option frequently, put it in the forefront. If you keep ssh'ing into a machine that's fingerprint keeps changing every time, just stop checking (unless it is actually hijacked, in which case alert). Start fstrim when it is actually needed (after a lot of deletion), not once a week randomly. Etc etc etc

An operating system based on something like the Pharo VM, Smalltalk VM or any similar mechanism where I can inspect all the code while it's running.

I know there have been projects like that in the past, but I don't know of any modern projects.

Yes, I accept the risks that come with that.

I want to combine a big database of data points about myself. Something similar to https://howisfelix.today/ but self-hosted, dockerized, easy to integrate and use. Plus import of the data from iPhone via Apple Health. Right now I can export it as a file but I don't have any means of parsing that file and importing it in to my own database.

edit: not sure why the data on that website has been blurred, but it's still available in HTML

I would actually love to make AI games. There's so much room for more player agency. We thought it might be something like some CYOA or AI Dungeon, but maybe it's closer to Balatro.

A network distributed block storage system that is as nearly fast and reliable as local storage. FYI, a block storage device can be used as virtual drives for virtual machines.

An offline tool that converts a checked out Facebook backup to a blog style website.

A modern replacement for Logisim Evolution.

Fork of Win 7.