What is a software you wish existed?

You might try the unexpected keyboard, unsure if it is completely comparable to the Hacker's keyboard, but it comes in clutch when I need those uncommon characters. I also like the way it works

Futo Keyboard is another one worth checking out. It also has a fully on-device speech-to-text engine, which is also available on its own as a separate app if you want to use it in combination with a different soft keyboard.

I am thoroughly enjoying HeliBoard.

I'm still using it on my Samsung S21, but I'm going to be very upset if it breaks on an update.

A Dvorak keyboard layout for Android with an intelligent autocorrect that implements a Pruning Radix Trie would be amazing. I use OpenBoard, but its recommendations for autocorrect are abysmal, likely because it doesn't use an English lexicon sorted by frequency.

FWIW automations can be built in HomeAssistant to do that, using their Sun integration.

For anybody interested in building out this function, here's a good starting point for that data. You could either implement a full calendar, or have it scrape the table data generator for a latitude/longitude.

Have you looked at Suntimes, and especially the Suntimes Alarms extension? I'm not sure of the total functionality, but it is based on setting alarms based on celestial events, including sun- and moonrise/set, along with others, I believe. I don't know if it has the capability of the "five minutes before" thing, but maybe worth a look if you haven't already.

Plus it's FOSS :)

An app to set different alarms for different days would be great. I couldn't find a good one on iOS.

Then you might want to look into Gemini, sadly likely no support for forums, but gives that smallweb indie vibe (because it is) without the jump scares and random [censored for sanity] of the early 2000s.

Metafilter.com is still out there living the Golden Age of the Internet.

The issue has always been economics. Either you make it ad supported or you paywall it. The problem with pay walling is that most of the young hip and interesting people aren't rich. So your platform becomes populated almost exclusively by older people who have disposable income.

I don't think you should rely "AI" on that 😅

Did you see that ludicrous display last night?

My go to is always to ask some kind of question to lead the conversation away from what making things uncomfortable.

I’ve thought about this recently and I think a topical conversation starter is usually all that’s needed to continue conversation on my own. Typically in an awkward situation both people just kind of stare dumbly waiting for the other person to spark a jumping off point.

If somehow you could incorporate news of the day/week and any other generic icebreakers (e.g. “see any good movies recently?”) you could do slightly better than this small talk calendar (which is admittedly geared towards college students).

Observational humor

Autohotkey

There is Autokey that is well documented and more powerful in many ways, but only works on X11 desktops.

Then there is ydotool that takes the desktop and turns into a headless browser basically, requires a lot more work for scripting but can do anything you can do on the desktop. Only supports Wayland.

Building the naming thing is precisely my pet project. I built one for my uni thesis. It was about 99% successful across about 10TB of media. But 99% successful is 100% useless. One day I'll return to this problem

A Linux distro or tool that does for all Windows applications what Steam did for Windows games.

Check out bottles, it is as easy to use/adjust as Steam's compatibility tools

FWIW I use GPT for this purpose and it is almost flawless.

I feel like Sonarr and Radarr might be of use to you. Even if you don't use it for downloading stuff via torrents you can still use them to import your libraries of media and auto rename them based on the formats you specify.

A robust and FOSS command line movie renamer similar to TVNamer with an option to rename folders while keeping files the same. The standard convention is Title (Year). Also, a reliable FOSS command line subtitle downloader which I can use with my Opensubtitles account.

Gods yes. The existing tools for this are a giant pain in the posterior. I'm stuck using TinyMediaManager at present because it's the least destructive option. I still have to do a manual touch up on things before merging them into the library. I think the best one I've ever used was Ember Media Manager but that's been a dead/defunct project for years now. In fact most of the projects in this space like GCstar and Video Hub are also dead. I think they all died about the time streaming services convinced people never to bother owning physical media or keep local copies ever again.

Closest thing I know of to what you're looking for is FileBot, it's also long in the tooth. There's a killer GUI tool for Windows called 1-4a-rename that has deep renaming capabilities, perfect for manually cleaning up a big mess in a hurry.

Frankly I'd settle for an operating system that can open a folder that has 50,000 media files in it without going into vapor lock for two minutes trying to collect all the metadata. Can I just get a damn list without trying to generate thumbnails and load twenty columns of metadata from the files please? I still have to fall back to midnight commander sometimes. I want that window open in ten seconds at most and I don't care if there are a million files in the folder, get better. Far as I can tell, every operating system still fails utterly at managing large numbers of files. By the time you hit 10k files it's lagville city. My music collection had to be sorted into artist folders from aa to zz just to get around this problem.

File systems suck. All of them, and the entire general concept of them as well. Why they haven't been replaced permanently by something more database driven (and thus vastly more performant) is still a mystery to me. /rant

Also, as for that last item... Ubuntu is rumored to be working on native windows app support. The .exe gets run under an emulation layer. Supposed to be simple as right-click and pick 'open under windows' to launch them. I'd love it, but I'll believe it when I see it, as this is a pretty spicy technical challenge to pull off. Right now one can do this using VMWare workstation, it just runs the apps in the windows VM and integrates their GUI elements with the host operating system. They seem and feel native, including clipboard integration.

command line movie renamer

have you seen mnamer? I've tried it before and it worked at least as well as TheRenamer used to.

I haven't used FileBot but I think at least some parts of it are open source: https://github.com/filebot

AutoHotKey

yep. AutoHotKey is a true gem. I recently converted a script to evdev and it is a lot more verbose:

RButton::Send {RButton}
  RButton & MButton::
    Send, ^w
    Sleep, 200
    return
  RButton & WheelUp::
    Send, {PGUP}
    Sleep, 10
    return
  RButton & WheelDown::
    Send, {PGDN}
    Sleep, 10
    return
return

evdev version: https://github.com/chapmanjacobd/computer/blob/main/bin/wheel.py

13 vs: 81 lines...

These are based on evdev so I think they could handle more advanced remapping:

• https://github.com/gvalkov/pyzmo
• https://github.com/moses-palmer/pynput
• https://github.com/acro5piano/wayremap
• https://github.com/joshgoebel/keyszer

Unfortunately, all of these programs require root access which is probably one of the big reasons that there isn't an equivalent to AutoHotKey on Linux. Doing anything interesting requires working around X, Wayland, etc but on Windows it seems like non-Admin users still have quite a bit of low-level hardware access

A Linux distro or tool that does for all Windows applications what Steam did for Windows games.

What does that mean?

This is appealing to me too. If I had to guess why it doesn't exist, it probably has something to do with the database work involved (lots of tables with lots of columns) as well as sourcing for all of the requisite metadata to make such a thing useful. Selecting a single type of media mitigates these issues to a significant degree.

That said, it's doable, but would probably need to be subscription-supported in some way (maybe only for advanced features) to be sustainable.

A site like that has been my dream for a long time: cross-media logging, cross-media recommendations, and an all-in-one wishlist for stuff you're interested in.

What games might you like if you enjoyed watching Black Mirror? What music might you like if you read cozy mysteries? A site like this could give interesting answers to those questions.

Another hard part of it though would be building a big enough userbase to have meaningful data to answer those questions and cover costs. I've seen quite a few indie logging/review sites come and go because they can't get enough people on them to be sustainable.

The lack of a thing like this is what lead me to build one of my own:

https://david.reviews/

Unlike all the other sites you mentioned, it's not a site where anyone can sign up and post. Its just a front end for my reviews database, which has movies, games, and books. That removed a bunch of headaches (storing login info, moderation, etc) while still giving me what I wanted (to be able to share my reviews).

Of course, there's also no community or algorithm recommendation features, but that's an ok trade off.

The whole front end is open source. The backend is airtable, but I've been meaning to publish my schema in a reuseable way (at least so people can produce similar results). All the data entry is automated too, so I just type a review and thoughts.

Anyway, all that is to say, a spreadsheet with some automation is probably best if you're just tracking for you! cc @ButteredToast

You may be interested in Todoist (see the "Original date or the completion date" section):

You can create a recurring task that repeats at regular intervals from either the original task date or from the task's completed date by using every or every!, respectively. Here's how:
Every: Completing a task that has a due date of every 3 months will set the task’s due date to 3 months from the task's original date. So if you created a task on January 10th with a due date of every 3 months, it will recur on Jan 10, Apr 10, July 10, etc regardless of when you complete the task.
Every!: When you complete a task with a due date of every! 3 months, it will set the next due date to 3 months after the day you completed the task. So if you completed the task on January 20th, the next occurrence of the task will be April 20th.

The "every!" notation sounds like it would do what you want. I use Todoist for everything that needs to get done, whether on a schedule or otherwise. I'd be lost without it.

On Android I use this feature everytime. You can create occuring event (every week, month, etc) and if you change latest event - like move a date forward or backward you will get question like 'do you want to change current event only or apply change to all events?'.

I haven't used it in ages, but I seem to recall that Remember the Milk does this. The next task is scheduled from the point when you mark the previous done. My only issue was that I couldn't mark things as being done on a previous day, if I hadn't gotten around to marking it done until the next day.

If you're on iOS, Sometime handles this use case quite well.

Question: why are you interested in this data, as a job searcher or investor or some other reason ?

Honestly this sounds like a great first project. Get some front-end library and a graphing plugin and find the data. If it is not there, you could use Python and Pandas/Polars to compile it yourself.

The job search site Otta pulls in company staff growth/decline percentage when you review a job posting. I assume you want something more broad, though.

There's a few sites that keep track of tech layoffs, but not sure if anyone is doing the opposite.

I saw a site with this exact data on hacker news a week or so ago. Let me see if I can find it.

Edit: I found the original site along with a few others.

https://layoffs.fyi
https://www.warntracker.com
https://intellizence.com/insights/layoff-downsizing/major-companies-that-announced-mass-layoffs

This is rather surprising. Aren’t most audiobooks just AAC files with embedded chapter metadata? That doesn’t seem like it’d pose too much of a challenge technically.

Wow, I've never even tried speeding up an audiobook, I miss too much stuff as it is. I just checked the main app I use, and it only does 2x. But if that's not a dealbreaker otherwise, I use Voice Audiobook Player (I use F-Droid, but linked the play store for ease).

It's pretty bare-bones, you drag your audiobook in a folder, which works well if you manage your collection elsewhere anyhow.

If you haven’t looked at audiobookshelf in a while, I would check it out again. It now supports running in windows as a single executable file. It literally could not be easier to set up.

Have you tried PlantUML? You write up text and it converts it into an SVG diagram for you.

I feel like text-based drawing tools are very convenient to edit. For example,

• https://swimlanes.io/
• https://arthursonzogni.com/Diagon/

Excalidraw is kinda magic in its ease and simplicity - just start using those hotkeys!
Code2Flow isn’t what you asked for but might be interesting to you.

I'm with you on this. I tinkered with something very basic along similar lines a couple of years ago, but as so many things I don't have the time/energy/will/drive to see it through.

The goal was slightly different, in that I was aiming for automatic layouting to minimize input, as I would primarily use it for trying to understand more complex connected networks of related items.

I also wasn't using NLP but just setting up a natural language syntax for it.

I got so far I could take input files like this:

A exists
B exists
A is connected to B
B is connected to C
D exists
D is connected to A
E is connected to A
D is connected to E
F is connected to E
G is connected to F

and automatically turn it into something like this: image

Differentiating implicitly and explicitly defined nodes was an important aspect for me at that time, which is why the nodes have different styling and why I declare them explicitly in the input.

You were able to configure your own set of instructions (here: "<x> exists" and "<x> is connected to <y>") and how the elements they affected would be styled (e.g. shape of node, type of line, arrows for directed connections, etc.)

As far as I remember though, with the tool I used for the layouting (graphviz dot language), having more explicit control on exactly how objects are placed relative to each other isn't really possible, or at least it's very much not trivial to automate.

One thing I did use it for was mocking up a flowchart for my understanding/interpretation of basically the state machine underlying stealth rules in D&D 5e, as a basis for discussing this with some of my players: image. Not the prettiest, and I'm sure it's not fully accurate to the rules, but it helped me reach a mutual understanding with the person mostly using stealth in that campaign as to how we would use it.

I love Mermaiddiagram for this. You write your diagrams in a declarative way and it generates a consistent version for you. I have a whole workflow using the CLI and Nodemon to have hot reloading diagrams.

Native is also cool because it’s a complete, fully offline binary, meaning that it can be run for the foreseeable future (even if this eventually requires hacks, compatibility layers, VMs, etc). Web apps are subject to sudden disappearance if the maintainer loses interest, no longer has the time/money, something happens to them, etc.

Native would likely be too temperamental and severely limit the potential of mass usage (and thus, community dev efforts) likely. Best case would be developing it in a native language, and focus on WASM builds (you can run WASM applications as native executables as well!), that way it can run on the web (or as an application) on desktop and mobile. I tried finding a project, but I can't seem to find anything? Closest is Bitmappery (great name), but doesn't seem as finished or comparable to Photopea, nor does it want to.

Edit (sort of, haven't even posted it yet): I found one, apparently I had even starred it on GH, but Graphite seems to be exactly what we (me?) are looking for. Seems to be incredibly comparable at first glance. They are releasing native builds later this year, but you can already use it in the browser as a PWA.

If you really must have it be native, then Gimp (yes, I know...) is actually really great and comparable when you enable the PhotoGIMP plugin.

I also tend to use Inkscape as a Photoshop more than anything these days, it's solid, and does what most people use Photoshop for (quick edits, and vector editing), it's most similar to Corel Draw or Illustrator (but certainly more Corel Draw).

I too have really wanted this. I wonder if this kind of app would get banned from the iOS store? I’m not sure explicitly what rule it would violate but it just seems like the sort of thing they wouldn’t allow.

I do this with their youtube streams. I run a cronjob every hour to pull with yt-dlp and then I have a php script (check my history) to generate the feed itself. It isn't perfect, but its good enough.

I would love a proper one.

With dynamic ad insertion, sponsor block as it exists on YouTube isn’t possible for podcasts. You would have to train some sort of ML model to try and identify sponsor segments.

Check out Picard -- it can auto tag if matching music can be found in the musicbrainz db

This is something I didn't knoe I needed!

Firefox devs are working on creating a better profiles function.

There is a auto container addon in Firefox which can automatically open each tab in a new container. Plus, you can force certain sites to always open in a certain tab if you want.

I played a bit with an eye movement tracking software that detects certain movements patterns, which people can then use to make to whatever. I'm sure it's way better these days for hand gestures and other accessibility/convenience tools

There are a lot of existing solutions for self hosted ecommerce (wasn't so long ago that was the only option) all with their own pros and cons. Just do a search for open source shopping cart. OpenCart is a pretty good option.

Smart idea to listen for trigger words. Not sure of an app that does that.

If you have iOS, you can set Overcast to skip the intro or outro by however long you’d like. Open the podcast’s main page>three dots> podcast settings>skip intro

It’s focused on enterprise, but I am pretty sure SmartSheets can do this. I have no idea if you can even get it as a normal consumer though.

So, just imagining this software, say you upload a PDF, get lots of comments. Then you make some revisions and re-upload it. What would happen to the existing sticky notes? Would you have access to the history? Would you have to manually mark each one as complete?