thevoidzero @ thevoidzero @lemmy.world

Posts

9

Comments

280

Joined

2 yr. ago

Someone already talked about the XY problem, so I'll say this.

Why sound notification instead of notification content? If your notification program (dunst in my case) have pattern matching or calling scripts based on patterns and the script has access to which app, notification title, contents etc. then it's just about calling something in your bash script.

And any time you wanna add that functionality to something else, add one more line with a different pattern or add a condition in your script. Comparing text is lot more reliable than audio.

Of course your use case could be completely different, so maybe give some examples of use case so people can give you different ways to solve that instead of just the one you're thinking of.

Yeah sure, I'll compile it in my OS. For any other OS, either I'm not knowledgeable about the tools available, and many of them that I am not going to spend money to acquire. If providing the binary a developer compiles for themselves would solve it, we'd not have that problem at all.

I specifically hate when program or libraries are only in compiled form, and then I get an error messages talking about an absolute path it has with some usernames I've never seen before, and no way to correct it as there's no code. Turns out when people pass compiled versions to the OS they don't use themselves they don't encounter the errors and think it works fine.

It's horizontal growth vs vertical. Growing by getting more coverage is growth but it's not sustainable growth that you get by getting the next generation into it.

I haven't tried many CAD softwares but AutoCAD has really intuitive UI. I used to be able to find most things by just thinking what tab it should be based on what it is. It actually inspired me to learn better programming and software design to make something intuitive. I haven't used it in years since I came to Linux so as long as they haven't changed it.

It's not even at OS level (unless you're thinking fonts are OS level). So basically you'd just need to define a font, with certain unicodes with images you want, and make sure it ranks higher than other fonts. So the application will search for any characters in that font, and then for those you didn't define, it'll search in other fonts.

I was thinking the same, smartphones would definitely do everything it can to make images sharp so it's probably not going to be easily stackable.

Still it feels like something should be there to combine the pictures to make better drawing, as there're softwares to generate 3D models from smartphone pictures.

Did you solve it? Recently there was a problem with graphics thing and downgrading mess from 1.24 to 1.23 helped me. It was in arch with AMD graphics, but some people said Nvidia ones also had the problem.

Edit: mesa not mess

Lol, that's a fun angle. They don't need all those fields coz they just get your information the other way

Similar. But I do contribute by adding things I want to some projects I use if it's simple enough.

And my pile of shit has like 40 stars, so maybe I have one or two other users besides me.

And how do I find a job like that?

Dam, this is actually awesome, I actually wanted to do something similar to this but couldn't figure things out.

I need to try again, might be able to make a virtual desktop screen and make xrandr pipe that output to that window.

Basically the thing I wanted to make is kind of like the picture in picture mode we have in browsers but for anything. Like just have those app open in a different virtual window and only remap the portion you wanna see into a always on top window.

Hi there, I did say it's easily doable, but I didn't have a script because I run things based on the image before OCR manually (like the negating the dark mode I tried in this script; when doing manually it's just one command as I know whether it's dark mode of not myself; similar for the threshold as well).

But here's a one I made for you:

 bash

    
#!/usr/bin/env bash

# imagemagic has a cute little command for importing screen into a file
import -colorspace gray /tmp/screenshot.png
mogrify /tmp/screenshot.png -color-threshold "100-200"
# extra magic to invert if the average pixel is dark
details=`convert /tmp/screenshot.png -resize 1x1 txt:-`
total=`echo $details | awk -F, '{print $4}'`
value=`echo $details | awk '{print $7}'`
darkness=$(( ${value#_(%_)} * 100 / $total ))
if (( $darkness < 50 )); then
   mogrify -negate /tmp/screenshot.png
fi

# now run the OCR
text=`tesseract /tmp/screenshot.png -`
echo $text | xclip -selection c
notify-send OCR-Screen "$text"

  

So the middle part is to accommodate images in dark mode. It negates it based on the threshold that you can change. Without that, you can just have import for screen capture, tesseract for running OCR. and optionally pipe it to xclip for clipboard or notify-send for notification.

In my use case, I have keybind to take a screenshot like this: import png:- | xclip -selection c -t image/png which gives me the cursor to select part of the screen and copies that to clipboard. I can save that as an image (through another bash script), or paste it directly to messenger applications. And when I need to do OCR, I just run tesseract in the terminal and copy the text from there.

Not for handwritten text, but for printed fonts, getting OCR is as easy as just making a box in screen with current technology. So I don't think we need AI things for that.

Personally I use tesseract. I have a simple bash script that when run let's me select a rectangle in screen, save that image and run OCR in a temp folder and copy that text to clipboard. Done.

Edit: for extra flavor you can also use notify-send to send that text over a notification so you know what the OCR produced without having to paste it.

You said you're not allowed to use a horse on freeway and it's not a fair comparison. But I think it is exactly that. Freeway is where the majority of traffic is and it's analogous to some of those major platforms where everyone is nowadays. You can use a horse and go to any place as long as there is land. It's just not practical to do it.

Yes you can make a website anyone can access but how will they find that website? You'll need to inform the people in the web, and that's dominated by those platforms. When people did the reddit blackout thing, reddit removed the posts and moving to lemmy, so without those posts we can't expect people to know about alternatives. There are probably so many websites that host contents for users to post and such, but how many have we heard about? How many can we find with an internet search?

Common knowledge doesn't mean people use it. It's easy to forget even if you studied about it in school.

For example you is singular and plural. But we rarely use you for multiple people nowadays, we just go "you guys", "you all", "all of you", or something else to disambiguate.

Languages move towards easy communication and simplicity.

About the malware thing. Won't the Linux use increasing in organizations give incentive for attackers to make malwares targeting linux? It's not like we're malware free, it's just that average user is informed enough and there is low use of linux making it not worth as much to target desktop users.

Thank you. I've been using grk4-rs so I went over their examples and the book. Looking at the c examples might be good as well. And I'll learn about the GObjects. So far I've been winging it based on examples I see and the properties of different classes.

Thank you, I'll go over them. I think I got a little bit of the basics now. But I need a little advanced knowledge. And most tutorials seems to be for gtk3 and many widgets are now deprecated.

Originally my plan was to make each slide edited separately when you right click and do edit. And be able to drag them up and down and everything, but my skills were not enough. After trying to make that for a week I gave up and made this simpler version. So listview and treeview models from that tutorial could give me more info.

Yeah, the bindings were a great help. I also tried tauri, iced, beavy, and a lot of libraries. At the end just using gtk4 seemed to be the best for my use case. I went through the grk4-rs book, and then started experimenting. I don't like GUI at all, but for simple cases it works well

Glade doesn't work for gtk4. There's cambanche, but it didn't work as smoothly as glade. Originally I tried just using glade and converting to gtk4 for a bit, didn't work great as some classes were different.

So at the end I went with just looking at gtk4 docs and writing xml on the editor. It'd be problematic once in a while if I miss closing some tags, other than that it went well.