No Stupid Questions (Developer Edition)

This is for models that use vertex colors. I want all faces with the max brightness of the object. So in daytime lighting it should look unshaded, but in a dark environment it should darken (like the thumbnail).

What I've tried:

• using vertex_lighting* and then setting the normal directly (eg NORMAL = vec3(0.0, 1.0, 0.0);)
  • this works, but only if the lighting is perfectly overhead... a slightly wrong angle or even too far into the middle of an omnilight will create a shadow
• attempting to save the original color value, change the normal, and then bring it the old brightness back via max() if the new value is dimmer
  • lots of moving parts/assumptions on my part here
• other math stuff with NORMAL/VERTEX, abs() and normalize() etc
  • on top of the direction failure, there's also failure to keep consistent brightness (with a spinning animation), and darkening across a TextMesh

I also know I could do this with unshaded and just multiply/mix it in the shader, but I don't really like the idea of making a custom lighting/detection system (rather than using actual lights, that is).

I am not totally against diffuse shading, something like a vertex toon shader (2-3 brightness bands) could be interesting aesthetically. If that's even possible, it seems like the vertex light calculation is hardcoded (though it's also possible I just don't understand it).

* this is a relatively re-added feature (4.4)

Seeking Direction: Sharing AI-Aided Knowledge Beyond Corporate Platforms

As a Reddit (and RIF) refugee, I'm searching for a space to share various useful information where it can be easily found later. The challenge is that the definition of "useful" varies among individuals, and I suspect that Reddit and other "for -profit" moderators are overly influenced by corporate interests.

In the past, I spent considerable time figuring out how to accomplish tasks unrelated to my primary goals—tasks I usually only needed to do once. Now, I tend to ask AI tools like ChatGPT, Perplexity, Deepseek, or other available models to assist with the task at hand. I've learned that by doing so, I can condense days or even weeks of effort into just a few hours, if not less.

However, a significant issue is that these AIs don't currently learn from my questions, and the knowledge they provide isn't propagated beyond the AI itself, unlike traditional sources. The AI models offering this information are essentially static, and the context of the current session is typically lost quickly.

The solution I envision is to take the results of successful chat sessions and post them somewhere on the internet where both humans and AIs can access and incorporate the information into their knowledge and training. The problem is that this approach doesn't align with the posting models of conventional for-profit services like Reddit or Facebook. I tested Reddit again this morning and immediately had a post rejected for "low effort," even though it would have taken me a long time to develop the content myself.

I'm looking for direction on:

Where to share this AI-generated knowledge so it's accessible and beneficial to others.

How to structure these posts to fit within the guidelines of non-commercial platforms like Lemmy.

What communities or instances would be receptive to such content.

Any suggestions or guidance would be greatly appreciated!

Like if I'm using print statements to test my code. Is it okay to leave stuff like that in there when "publishing" the app/program?

Edit: So I meant logging. Not "tests". Using console.log to see if the code is flowing properly. I'll study up on debugging. Also, based on what I managed to grasp from your very helpful comments, it is not okay to do, but the severity of how much of an issue it is depends on the context? Either that or it's completely avoidable in the first place if I just use "automated testing" or "loggers".

To what extent will these changes, and potential future changes along the same trajectory, affect the use of un-substitutable apps on Graphene OS?

Play Integrity API - 2024 Dec

The transition to the new verdicts will reduce the device signals that need to be collected and evaluated on Google servers by ~90% and our testing indicates verdict latency can improve by up to ~80%.

the huge ratio of reduction suggests to me that the attestation is being offloaded from Google servers to on-device AI, but maybe i assume wrong. my instinct tells me Google would always make this impossible for 3rd party OS to implement anyway.

Hypothetically, If implementing that AI in Graphene could allow most attestation-requiring apps to install and run normally, is that something the Graphene devs would do? i know it would have to be secure and private, so assuming there was a way...

I'm not pro AI, not the surveillance big data capitalism kind at least. just wondering about the scenarios and what prep or extra work i would have to do.

In my understanding, the options need to be customized for each machine, as well as the fact that making packages for tons of distros can be a lot of work for solo or small team devs, and that's why some software is provided as .tar only.

but it seems like the install process on the user side could be automated to a single command or drag-drop, as long as the script would throw informative alerts for any errors and the user is prepared to take over manually.

does something like this exist in a standalone form that's not bundled like snaps or flatpaks?

if not, is there a broadly-applicable reason (security, damaging OS, etc) that makes it a terrible idea? or simply that no one has gotten to it?

~~The idea is making deployment so simple even a 13 year old could republish the website in cloudflare or github pages.~~

I'll admit it, this wall text was very dumb!

Any other tactics are welcome

To my understanding:

Many terminals are capable of displaying multiple fonts at the same time, say latin unicode characters in font foo and japanese unicode characters in font baz. In urxvt at least, it is also possible to have one font in a certain size and the next font in another size. However, no font can have a size bigger than the base size, the size of a terminal cell.

Why is it not possible to have multiple terminal cell sizes? For exampleso one line has terminal size 8 and the next line has terminal size 12.

The wikipedia articles are terribly written (for math loves or people who just need to refresh their knowledge).

What is a "sum" of types? What is a product of types? Is it possible to Cat x Dog or Cat + Dog? What does that even mean?

So I want to build blender fork but it fails to build on Visual Studio 2022. There is already a patch and a open PR that fixes the issue.

I have already git cloned the repository and I would want to only get the patch into my local repository. So I can build from that.

Limitation of using drag and drop Images in readme.md?

One I am aware of is the size limit that no image size should be >10 MB. Are there any other limitations when using this (for example: retention period, storage capacity, etc)? I want to link those images outside Github.

I am aware of uploading images to the repository and linking by

[image](./path/to/image/image.png)

You see this with some apps (I think ReVanced is a popular example?) and games occasionally, and I've never been clear on how they do it.

Hello. I have Windows - Ubuntu dual boot and I'm trying to move space from Windows to Ubuntu. I've already freed space from the Windows side

I'm pretty sure that I've read online that it can be dangerous to move the unallocated partition, because next boot to windows can corrupt my Ubuntu system. Is it true? Also, when I'm trying to move the unallocated partition, there's no option to "move/resize", so I swap them with the next following partition one by one. Is it the right way to do it?

So I'm a baby dev, still in Uni and they don't allow internships in 4th year due to some issues with it so not even that exp wise.

I don't know enough, and I'm trying to learn but there's so much! My Uni degree doesn't cover security at all. Which is shit, bc I think I want to work in that? Mostly I'm just spooked and want to understand everything I can 'cause I love the internet and want to feel safer wandering about it.

I'm scared of clicking on links. Even ones here, like there was a post about a book list earlier and I was just there like "Cmoon.... someone please have posted the lissssst."

Would anyone be willing to share what they do for their own security? Especially if it's ridiculously over the top. Included reasonings and details would be adored!

Also, if anyone has any books or references that might be good for learning sec from a programmatic view rather than a IT view I'd really love that! Anything at all.

Regardless, hope anyone reading this has an absolutely wonderful day and best of luck with everything you're up to!

Can't locally download 41 GB loaded image which is provided to replicate GitHub action locally but don't want to commit and check every time too, is there any third option?

I think from what I've read that this is the case, but I've read some other info that's made it less clear to me.

On the second part of the question regarding container engines, I'm pretty sure that may also be correct, and it kinda makes me wonder a little about risks of engine lock-in, but that may be a little out of scope.

cross-posted from: https://programming.dev/post/6513133

Short explanation of the title: imagine you have a legacy mudball codebase in which most service methods are usually querying the database (through EF), modifying some data and then saving it in at the end of the method.

This code is hard to debug, impossible to write unit tests for and generally performs badly because developers often make unoptimized or redundant db hits in these methods.

What I've started doing is to often make all the data loads before the method call, put it in a generic cache class (it's mostly dictionaries internally), and then use that as a parameter or a member variable for the method - everything in the method then gets or saves the data to that cache, its not allowed to do db hits on its own anymore.

I can now also unit test this code as long as I manually fill the cache with test data beforehand. I just need to make sure that i actually preload everything in advance (which is not always possible) so I have it ready when I need it in the method.

Is this good practice? Is there a name for it, whether it's a pattern or an anti-pattern? I'm tempted to say that this is just a janky repository pattern but it seems different since it's more about how you time and cache data loads for that method individually, rather than overall implementation of data access across the app.

In either case, I'd like to learn either how to improve it, or how to replace it.

What are some alternatives to browsing instagram "outside" of the proper app?

I put nitter.net and invidious as examples, as they allow you to browse Twitter and Youtube "outside" the official site/app and without being tracked. Instagram hates me trying to look at anything from the computer.

I need an open-source AI front-end (use with openai api key) with following option:

• Do web search
• Chat with document as context
• More Plugin

Please suggest me some open-source project that work.

I have 5 years experience as full time job.

However I never do freelancer job before.

How do I start on finding freelancer job ?

It is a common sentiment that managing dependencies is always a big issue in software development and the reason why so many apps come pre-bundled with all the requirements so it reliably works on every machine.

However, I don't actually understand why is that an issue and why people generally bash npm and the way it's done there. Isn't it the simplest and most practical solution to a problem - you have a file which defines which other libraries you need, which version, and then with one command you can install them and run the program?

Furthermore, those libraries and their specific versions can be stored elsewhere and shared across all apps on a system so you can easily reuse them instead of having to redownload for each program individually.

I must be missing something since if it were that easy, people would have solved it years ago and agreed on a standardized best way, so I'm wondering what is the actual issue and a cause of so many headaches.

In CSS, let's talk about srcset or image-set. In that context, you can define which image the browser loads using 1x, 2x, 3x, etc. These refer to pixel density. (In the case of srcset, you can use pixel dimensions too, which sidesteps the issue I'm going to talk about, but it still occurs in image-set, and also is still weird to me in srcset, even if you can side step it.)

So, assuming, say, a 20" monitor with 1080p resolution is 1x, then a 10" screen with 1080p would be, technically, 2x - though, in the real world, it's more like a 6" screen has a 1000x2500 resolution - so, I don't care about math, that's somewhere between 2x and 3x.

Let's imagine a set of images presented like this:

srcset(image_1000x666.webp 1x,
image_1500x1000.webp 2x,
image_3000x2000.webp 3x)

then an iphone 14 max (a 6"-ish screen with a 1000x2500-ish resolution, for a 2-3x pixel density), would load the 3000x2000 image, but my 27", 1440p monitor would load the 1000x666px image.

It seems intuitively backwards - but I've confirmed it - according to MDN, 1x = smaller image, 3x = larger image.

But as I understand it, an iphone 14 acts as if its a 300x800 screen - using the concept of "points" instead of pixels - which, in the context of "1x" image size makes a lot of sense - but the browser isn't reading that, all it seems to care about is how many pixels are in an inch.

I made a little page to demonstrate the issue, tho I acknowledge it's not hugely helpful, since, other than using your actual eyeballs, it's hard to tell which image is loaded in the scrset example, but take a look if you want.

https://germyparker.github.io/image-srcset-example/

After some time of practicing PyQt I found out about Kivy, as a beginner which one can be considered with an "easier" syntax to learn? After several days of practicing PyQt I feel I would be throwing away my learning for Kivy.

PyQt is not that complex really but what I hate with all my heart is using Qt Designer and then exporting that to a .py file, I really don't like the way it formats the text, but that's just a personal detail. I know I can perfectly do it manually but all the tutorials and I mean ALL of them always do it in that way.

I want to clarify that I'm interested in Android application development, that's really a strong point for me of Kivy.

I've been learning Python by myself for about 3 years now and I can say that I know quite a lot but I don't really feel confident in my own programming skills and always after a while of practicing or reviewing I end up quitting because I feel exactly this.

I don't know how to explain it, but I really feel like I'm in a cycle repeating the same noob exercises over and over again.

For example, lately I have been practicing a lot PyQt but I really feel that I am wasting time when I don't learn a new concept or I don't memorize something and I need to look at my notes to remember how to do it, and also that practicing with online courses, especially with Youtube is often a challenge because the authors do things differently and I get confused by that. And when I want to learn something new the amount of information overwhelms me and I feel tired because of that.

As a Linux user I know that what I just said is stupid, because for example it is impossible to learn all the commands in the world, you just really learn the ones you use most regularly but in programming I feel that for example asking ChatGPT (or any ChatBot) counts as cheating for some reason, I don't know how to explain it.

I really consider this probably a mentality problem more than a skill problem because honestly even though I know I can I don't feel sure how to program, many times I even doubt the name of my variables or my functions.

Thanks for reading my silly post!

Note that I'm using autohotkey v2, not v1.

I want to run two different autohotkey scripts. I want to trigger a hotstring in the first script, the output of which ends up being part of the hotstring trigger for the second script. Is this possible?

Here's a simplified version of my intended workflow.

Script 1:

#Hotstring EndChars \
#Hotstring o
#Hotstring ?
::iv::ǐ
::av::ǎ

Script 2:

#Hotstring EndChars \
#Hotstring o
#Hotstring ?
::nǐ::你
::hǎo::好

So the idea is that I can type niv\ and the first script will convert it to nǐ - then I can immediately type \ and the second script will convert it to 你. So I type niv\\ and my text goes from niv to nǐ to 你. I can then type hav\o\ and have my text go: h, ha, hav, hǎ, hǎo, 好. So I can do niv\ hav\o and get nǐ hǎo, or I can do niv\\ hav\o\ and get 你 好. Both writing systems in a reasonably simple format.

There are reasons I want to set it up like this. The first script has dozens of functions beyond writing in pinyin/chinese, and I share it with another person - so I don't want to add potentially hundreds of random Chinese hotstrings to it, just the special pinyin characters. That's why I'm using two scripts.

But I also realize I could just make "niv" and "havo" their own hotstrings which go directly to 你 and 好 without the intermediate nǐ and hǎo. I don't want to do this mostly because I think the system I have in mind is prettier - type it correctly in pinyin first, then have it correctly convert to Chinese.

All of that aside: I've gathered that this is probably possible using some combination of SendLevel and #InputLevel - but I've tried a bunch of different combinations and ideas with it, and haven't successfully had one script trigger another yet. Even in simplified toy scripts, which is a little discouraging. Ideally I'd be able to do this with as few changes to the main script I share with another person as possible - the script that handles the Chinese can be as complicated as it needs to be though. Anyone know how to make this work?