This thread is for anyone working on personal projects to share their progress, and hold themselves somewhat accountable to a group of peers.
Post your project, your progress from last week, and what you hope to accomplish this week.
If you want to be pinged with a reminder asking about your project, let me know, and I'll harass you each week until you cancel the service
Jump in the discussion.
No email address required.
Notes -
Christmas turned out to be a crunch of it's own, but I managed to wrap up the script for migrating data from V1. If I manage to eek out some time I'll deploy the latest versions as it has some bugfixes, QOL improvements I'm eager to get my hands on in my daily usage.
How are you doing @Southkraut?
Decently, thanks for asking. Got some time in, but I was admittedly very sloppy.
Got some UI work done, basic but it works. Layout, buttons, nothing fancy.
But mostly I dug into procedural geometry. My old hobby horse. I've implemented some for every engine I ever worked with, often several times over. This time I decided to do it differently, to not make the same old mistakes again, and just let ChatGPT do it for me. The results were decidedly nonfunctional, and I had to smooth it all over by hand quite a bit just to get it working at all. That said, the good chatbot did have some input I hadn't heard before, and it did help me get it running in Unreal C++, and it was useful for spotting obvious bugs that I couldn't be assed to sift from the logs. But overall I am indeed just re-doing what I had already done previously.
It's supposed to give me roughly spherical planets to deform and paint. Right now all I have is an icosahedron with vertex coloring. Subdividing it for greater resolution leads to incorrect results. That'll be the next thing to tackle - find out why some triangles are missing after subdivision. Shouldn't be too hard, but right now I wish I had just done it all myself instead of accepting LLM muck. Maybe it would've turned out better if the LLM had been integrated into my IDE, but I use JetBrains Rider, and their integrated AI assistants all require credit cards even just for the trial versions. I do not own a credit card, and I decided against finding a workaround like a virtual credit card. Maybe I should switch to Visual Studio and get Github Copilot - it's worthless at work, but maybe it'll do the job for Unreal development. Maybe I should trust my instincts and just do it all manually. Maybe just asking ChatGPT now and then is the optimal solution. I have no idea.
More options
Context Copy link
More options
Context Copy link
So I've been grinding leetcode to prepare for a technical interview. We'll see how it goes.
The company sent me this "preparing for a technical interview" packet with a bunch of articles saying a good amount of time to prepare is 2-3 months, and here's 150 problems you can do to study.
I have less than two weeks.
So anyways, I'm plowing through the problems about 10-15 a day, trying to get the most out of them I can. Come up with a pseudo code algorithm, check the hints to see if I'm on the right track, do my solution, check it against what they came up with. Some problems I can solve off the dome without a debugger or anything. Some take a little more work. Some I'm nowhere close. I'll probably circle back to those if I have time at the end of all this.
Probably about 1/2 (so far) I still remember from college 20 years ago, although it was rusty. Some of it I'd never seen before in my life and never would have gotten it in a million years if I hadn't just been told "you solve problem X with solution Y". Like Floyd's cycle detection. Some I just definitely need more practice with so I can belt out solutions with less fumbling, like variable sliding windows. My time in the c#, stitching libraries together ghetto have not served me well.
It actually reminds me a lot of when I took Differential Equations. At least the way the class went for me, some Chinese guy just gave us very specific solutions to very specific problems, and then our job was to torture the equation's we were given into the form of one of those very specific problem:solution sets. And so it goes with leetcode. You just need to memorize the the problem:solution sets, and recognize that what they are asking is actually problem y with solution y.
At least that's how it appears at the moment.
Then again, supposedly the bar is hell. They asked me a few extremely basic technical questions in the screening interview, and then I was told I was the only person who got them all right. A buddy of mine who's hiring (but is not allowed to hire in the US) gave me a mock interview and basically told me even basic competence is rare, and the technical interview for a normal company is basically testing to make sure you aren't a complete fraud. Some of the stories he told me about fraud in the hiring process blew my mind with their brazenness.
I've been on both sides of interviewing. For better or for worse leetcode style stuff has become the standardized testing of hiring. Generally better since you can't really grind leetcode enough and still pass tech interviews if you have a <100 IQ. But it also rewards smart dorks who mercilessly study it but are otherwise shit developers.
You definitely lose people sometimes because they brainfart during one of them, or they get anxious and underperform. More experienced interviewers know this and get a sense fairly quickly if you're going to actually get the right answer if you have more time and lower pressure. The best companies use these very quantitative measures as foils to have more qualitative deep technical conversations and see how well you work together. People get hired all of the time even though they thought they failed the coding interview since they didn't get the right answer or they only wrote out an O(N^2) solution but were clearly on the right track in discussions for an O(N) one and they thought you were cool and reasonable.
Maddeningly, you never truly know why you fail most interviews. Try not to read too much into it if you do.
Yes, there's an adverse selection problem. The people looking for jobs tend to be terrible. (Also if a job opening exists, it probably sucks: good people fleeing a shit show or because incompetent hires were fired.)
More options
Context Copy link
More options
Context Copy link