What are you doing this week?
15 points by caius
15 points by caius
What are you doing this week? Feel free to share!
Keep in mind it’s OK to do nothing at all, too.
My wife and I continue to work on https://uruky.com, a simpler Kagi alternative, based in the EU.
Last month we reached 200 monthly active accounts (we’ve passed 250 now), and last week we launched support for XMR/Monero payments via ProxyStore.
Feature-wise, for July we’ve already shipped a lot of visible and less visible things. We’re currently looking into increasing our own index, focused on indie/small web, and plan to add a couple of new search providers in the upcoming weeks.
Doing my best to relax while visiting parents, and learn a new song on mom’s old classical guitar.
It helps that I can’t log in to work GitHub or Slack from any of the devices I brought with me. (This is by design.)
I'm finally working on getting a homelab set up! Since it's just a hobby project and I don't have to worry so much about timelines I'm overengineering the heck out of it and trying to do everything with nix.
Studying Medea, a Greek tragedy where a foreigner, smart and capable woman that has been brutally cast aside by an entire society and her husband decides to break the system by doing things unheard of ever since (or after) … and as a result the Gods save her instead of punishing her (which is extremely provocative take… On the other hand Greek Gods are cruel, evil, just at times… there is no point trying to reason with them).
I saw the play on Saturday and was shocked by how good the leading actress (Kariofilia Karabetty) was at a virtually impossibly difficult role. This is life theatre not TV. She had three thousand ppl, well versed at theatre, on their toes for three hours. It was a surreal experience (the moonlight, the ancient theater and the scenic views help ofc). I’ve seen this play before but it was emotionally plain, manageable… Saturday’s version was a masterclass in emotional swings.
So now I’n reading Nikita Milivojević take on the original script, through the play’s book and it’s fascinating. I will probably write a blog post about it when time allows.
Work: migrating data from EBS to EFS, working on an automated rotation script, exploring a new monitoring provider…
working on my build system for C in C called mate.h, trying to add a few more features and cleaning up code
Building a VST. I really like what Gullfoss does when mastering music - it seems to make basically anything I put it on sound more clear. However, I'm not a fan at all of iLok, the obnoxiously anti-consumer DRM solution the plugin uses. So I'm studying psychoacoustics and trying to understand the effect so that I can reproduce it in a form that isn't encumbered by DRM.
What's your current understanding of the general principles behind how Gullfoss works? This is the first I've heard of it, and I probably won't try it, but I wonder if there are any general mixing/mastering lessons to be learned here.
It's a dynamic EQ mixed with a dynamic frequency unmasker. Specific frequencies that are too loud will make surrounding frequencies inaudible, as a function of their level and how close together the two frequencies are. There are some attempts to quantify this like the Bark scale. Gullfoss is to some extent attenuating the dominating frequencies and then boosting the frequencies they were masking in order to make everything more intelligible.
There are other plugins like Neutron or Trackspacer that do something similar but they require sidechaining - Gullfoss seems unique in that it can be used directly on master.
My understanding is that you could accomplish the same thing with careful use of dynamic EQ while mixing, but plenty of good engineers say that this speeds things up for them and improves quality over what they can do without it.
currently doing my best to not implode while assembling my meltybrain style (basically a beyblade that spins at 4k rpm) combat robot in time for opensauce. shipping times are killing me atm
I'm working on some tweaks for my word guessing game https://wordbattle.fun - You play against family/colleagues/friends in teams.
It tracks your solves and some other fun metrics. Totally free, no ads or other crap. No login needed to play.
Continuing to work on a RV32I CPU emulator in ProtoFlux (yes the funny node-based VR programming language)
I'm finishing up Zicsr and next up is interrupts and some first memory-mapped devices, plan being to get embassy working on it :3
I have combined a plethora of open Chinese datasets to make a quite comprehensive Chinese learning thing (just Standard Mandarin; no other dialects).
Could be fun. It’s honestly enormous but I’m hoping it’s organized in such a way as to be accessible.
Setting up a honeypot for obnoxious web scrapers!
Background for this is I'm adding ML support to Heavy, my attempt at a less intrusive web app firewall. Proof-of-work is annoying, so I want to see if I can use classical ML techniques as a pre-filter. But to validate different approaches for doing that, I need a variety of scraper traffic—hence the honeypot.
Git forges are the stereotypical thing that overzealous scrapers like to knock down, so I spun up a Gitea container and exposed it to the web. As part of that, I've been brushing up on container technologies, specifically Quadlets. I've bounced off Docker a couple of times in the past, but on this project I've been having a good experience with containers so far.
I'm going to set out more "bait" than just that—if any of you folks have suggestions, I'm all ears!
I've had a good experience with Altcha (MIT, self-hosted PoW captcha).
As for more ideas, I'd suggest setting up some kind of Nextcloud instance (and don't forget to setup a robots.txt that blocks everything everywhere, to "only" catch the "bad guys").
Would be keen to know how this works. Maybe the techniques can be useful beyond the product you are building.
Continuing to learn OS fundamentals by writing baremetal Rust for my Raspberry Pi. I wrote a general-purpose memory allocator and it was broken, so I used proptest-state-machine to find the flaws. The tests pass now, so it's time to boot up the hardware again. Once I've got that working, I should be able to continue the USB device enumeration process to eventually detect a keyboard or mouse.
Working on onboarding flow and puzzle RAG quality pre-launch for my daily programming puzzle site @ https://dailyprog.club
Also might work on an easter egg for my resume https://codeberg.org/lvmbdv/resume
I am trying to run (/learn) hermes via podman. Using it to import all my bank statements and log them in hledger format.
Moving. Trying to find work.
Say what activities do people find helpful while between jobs?
Personal: I continue working on Abject and recently published a post which explains some of my reasoning behind the project (https://blog.mempko.com/a-love-letter-to-object-orientation/) . This week I am working on robustness, especially in the UI so that users can generate more interesting things.
Work: I am working on Thetix version 1.5 of our financial AI harness. Includes performance, quality, and robustness improvements. Since so many SOTA models are coming out we have to re-run our evals to continue making sure we are the top (https://thetix.ai). Then I will switch to the more crazy Thetix 2.0 work that brings over a lot of great ideas I've been experimenting with Abject. I want Thetix to be so good there is no other obvious choice for a financial/investing agent.