Mapping software that can give directions the way human navigator would.
When I’m driving in my own city, my mapping software should be intelligent enough to know that I am aware of most of the roads; it can track me.
I don’t need to hear
“Keep straight on Highway 101 West signs for Highway 101 West for 300m, then take exit 104 South signs for Highway 104 South, take exit 104A South signs for 104 South, merge onto Highway 104 South signs for 104 South. Go straight on Highway 104 South for 400m then take the left lane and turn left on route 40 Eastbound signs for route 40. Boodle-ding you are on the correct route. In 200m turn left onto route 40 Eastbound signs for route 40 Eastbound. Turn left onto route 40 Eastbound signs for route 40 Eastbound.”
… When what is needed in a realistic sense is the following:
“In 300m take the exit to 104 Southbound then after 400 m, turn left at the first set of lights onto route 40”
That would be a meaningful improvement. I moved - basically sight unseen - a year ago to a new town. Day one I needed every bit of turn by turn. Now, if I’m headed to any of the four or five places I bother to go, I just set up the map as a CYA and a simple “yep, make that turn you’re planning on” would be sufficient.
Then there’s a 90 minute trip i make every two weeks, that I know fairly well but not like I’d know a daily drive. The first hour is “Jump on 74 west, take exit for 57 south, and go a ways”. That part I have down cold obv.
After I get onto hwy 36 tho, damned if I can remember where the (poorly marked) left turn onto CR 1300 is.
Better still would be an adaptive mode. Leave me tf alone with my CCR playlist until I’m within a couple miles of that poorly marked turn. THEN help me out with a gentle reminder.
The hour or so of instructions prior to that point are wasted and would be pretty easy for AI to figure out I don’t need help on that part.
Sadly, that description was motivated by a direct experience I had a few days ago with Google maps.
But I’ve definitely had that other experience as well, where the next leg of the trip is to go straight for 20 minutes, and Google Maps chimes-in every 30 seconds to remind me to stay straight.
The most recent meaningful upgrade I have seen in the software is that instead of deciding to either play each piece of audio through the phone speaker or the Bluetooth, quite at random, it will now play things through the Bluetooth. Livin’ in 2024.
It would be really nice to have a parametric 3D modeling software solution that was on the same level as Blender or KiCAD. Every time I try FreeCAD I end up moping out pretty quickly.
FreeCAD in its vanilla state (without tweaks) is pain in the butt in term of usability, it’s still bit “hard” to recommend for casual user when they are coming from commercial like Autodesk Inventor, Solidworks and such.
Recently Ondsel Team created sort of modules for FreeCAD which also marketed as standalone product that lets you to make it functions like Fusion 360 in term of cloud connectivity, their free tier also good even better than Fusion 360 Hobbyist License. They also contribute toward FreeCAD upstream for some general improvement though some of them are exclusive on their own implementation.
The day FreeCAD 1.0 dropped, I definitely going to try it out because I’m stuck on Fusion (needed for collaboratory work) ever since graduated from college.
Yeah, I was taught using NX and coincidentally also use it at my current job, but I have dabbled in almost all of the other commercially available platforms and so far FreeCAD took the longest for me to “pickup” as it were.
You are probably already aware but there is OpenSCAD which allows you to model via programming rather than by UI. Not really an apples to apples equivalent but I find it decently interesting. I do wish there was something a little more more overtly friendly to beginners like Fusion360 though.
To be honest, it is first time I discover OpenSCAD, probably due to its nature modeling by programming rather than visually.
Logically you do want to model by visually especially when it’s more complex geometry and perhaps that’s reason why you may seeing them less getting recommended in general when something like BricsCAD (Education license), OnShape exists.
I agree Fusion 360 (on Windows) with Free Hobbyist/Personal license is good start to learn CAD modelling for free though as year went by the Hobbyist license becoming stricter and limited in term of policy which raising red flag for hobbyist, not to mention Autodesk also converting users lifetime license into subscription without any notice is enough reason to stay away unless your job provides you those CAD program licenses.
Well in which case you probably won’t believe me but I’ve been using Kagi for a while and am extremely happy with it… no link with the company at all other than a very satisfied user 🤷🏻♂️
You can improve Google a bit using ublacklist but it still wasn’t anywhere as good as Kagi.
I will say Kagi isn’t as good when it comes to looking for local businesses or services… I still use Google for that but you can do that within Kagi with a !g and it anonymises the search
Matrix will never replace discord, because implementing gifs took them like 10 years so far and it’s still not even a concept. And imagine discord like experience without gifs. Impossible.
Which is sad, because Matrix is otherwise absolutely great! It’s just not focused on casual users at all.
Well, majority of people discord is focused on do use them. That is how it is. When Matrix offer something meaningless (for them) like e2ee or federation in exchange for not having gifs and being “too complicated to set up”, you’ll never get this people to use it. It just makes no sense for them.
Hear me out about GIMP - there’s this patch that re-arranges it to be as close to Photoshop as possible: github.com/Diolinux/PhotoGIMPOf course it’s not 100% the same, but I think you should give it a try.
I know it doesn’t tick the boxes but there technicality is a Microsoft made, open source, alternative to Explorer. The original File Manager for Windows 3.1 and it is still seeing active development. Just thought to bring it to attention for a bit of nostalgic fun but I actually find myself using it a fair bit.
vscode extension that summarizes Java’s exceptions for me so that I can easily find what’s wrong and which line caused it instead of scrolling through 100 lines of exception/error stack.
Something that can utilize a Shazam API or something similar and go through my entire music library (which is full of hundreds of tracks named “Track 1”, “Track 2”, etc.) and title them appropriately, ideally with correct metadata and album art. I would pay a lot for this.
Or whenever using your credit card online. The pro version would be that it turns off functions successively depending on your BAC. At some point the only unblocked function would be to call a cab to go home.
Not so much software as an element that needs to be part of so much software but, for some unfathomable reason, is not:
Picklists which work properly. If I type “U”, do not give me a screen full of Ts with the first U at the bottom. And let me type more than one letter to get to exactly where I need to be in the list. It can be done but it rarely is done and it does my head in.
Also, if you must include a scroll wheel to enter numbers, make it possible to just fucking type the numbers instead.
This is probably not what you intended this thread for but I’ll take any chance I get to ask UI designers to get a fucking grip…
Software developers who never have, and never will have to, use the software for real. I think every coder should be forced to use their own software for one month out of every year they work on it (and be able to do the job that goes with it because how the fuck else are they going to get a clue?).
And fucking stop making PC software that looks like it was designed to be used on a phone. I cannot do my job on a phone, no one would ever do my job on a phone, everyone who does my job has at least two large screens. We do not want to click a million times to do one simple task, and we do want to be able to see masses of information at the same time.
I think every coder should be forced to use their own software for one month out of every year they work on it (and be able to do the job that goes with it because how the fuck else are they going to get a clue?).
At my company, all developers spend time every year in customer support. It gives us first hand experience with what our customers are running into and asking for. We also work directly with field consultants on their projects. It’s not exactly this, but it’s pretty close, and it works really well.
That must help a lot. But often, when I am quietly cursing them, I just want to make them shadow me for a day to see, and feel, the impact of their ridiculous decisions. We have given them written explanations, had meetings, shown screenshots. But nothing gets through. If they had to spend a day a week using it, they might actually do something about it.
Software developers who never have, and never will have to, use the software for real.
Yes. The customer doesn’t necessarily know what’s possible or know how to articulate what features they want. I spent one week in a position where I was using my own software for production and immediately made several simple enhancements once I had hands on experience with the expected business process.
Every programmer should go through an exercise like this at least once in a while.
I am quite old, so remember the transition from scientists writing their own software to systems analysts who specialised in writing software that was fit for purpose. And that was exactly the ideal: the systems analyst was supposed to be someone who could code as well as their programmers could and understand the job the software was designed for as well as the customer did.
None of that seems to have happened. Some of the kids who could code got lucky with billion dollar jackpots from very low hanging fruit. And ever since, we’ve just been hit by waves of kids who can code going straight into software development with absolutely no experience of how work works.
It’s a difficult problem to solve. I have an aunt who developed software in the '60s and '70s who had to retire early because the languages she used became obsolete (apart from a brief bounce running up to Y2K). But it is a problem we absolutely have to solve. So much shitty software, wasting so much effort, for the developers and users alike.
Here’s a basic UI thing that needs to happen: spatially stable navigation.
When I scroll up, something should not appear unless it was just hidden by my scrolling down.
When I hit “back”, I should always be where I just was.
These are, in some ways, the same thing. Scrolling up on a webpage is, quite often, intended as an “undo” for the previous scrolling-down action. When I scroll up, I want to see the last thing that disappeared under the upper fold. I don’t want to see your menu, which wasn’t there before.
A templatable OCR app that maps areas or shapes to excel fields.
If you have a product tag with different serial numbers or product details and a standard layout it would be really useful to be able to scan for a tag shape, apply an overlay with each block of relevant data and then map that block to a cell address.
Take photo of product tag x100 OCR and edge find on product tag Select/draw areas Assign areas to spreadsheet cell or column. Apply and check with second photo. Confirm function and process next 97 images automatically.
Thought of it for work but would be great for food labels and nutrition information collation as well. All sorts of paper->digital stuff.
There have been attempts at this with store receipts, so you quickly scan your groceries into your budget or inventory app. That has been a difficult problem, in part because stores change receipts so much.
I’ve seen and used the on the fly versions for expensing receipts and such but I’ve run into standardized tags at work a few times where it would be nice.
Add comment