Pins

I found that I had Structured data has schema.org validation error on essentially every single page on my blog, turns out I had made some changes and have never tried to validate it. Damn json and its hatred towards trailing commas.

Long live RSS! Rss is not dead David, you are right there. I really agree with David that learning a topic well enough to form thoughts and write about it really help learning. You don’t need to be an expert, but forming your own thoughts, putting ideas in words takes a lot more than surface level knowledge. When you try to write or speak about something you quickly realize where your holes in understanding are. Blogging helps me learn. When I commit knowledge to writing it reinforces what I know and shines a spotlight on what I don’t. Most topics require additional research. Even then, I occasionally get things wrong, or miss different ways of thinking, and I welcome corrections. I’ll often update and enrich my posts based on feedback. Without my blog I’d miss other points of view. As they say, the best way to get an answer on the internet is not to pose a question, but to assert the wrong solution! Most feedback I get is constructive. Sometimes it’s blunt but I try not to read into unspoken sentiment. Some people are more direct. If the end result is positive learning, I can take a hit or two.

In fixing a bunch of meta tags, I introduced Open Graph URL not matching canonical on every page by having trailing / on canonical and not on the og:url. This commit will fix the error.

I like Davids idea for cotton coder here, reminds me a lot of Thoughts, which turns out to be mroe commonly called a linkblog. I can relate to David heavily on gathering too many side projects and soem collecting more digital dust than you would really like them to. I use thoughts for quick publishing, very similar to David’s notes. I have tags and titles, but the titles are a reflection of the post I’m taking a note on. They are short and sweet, I put just enough thought into them without overthinking them. They live as a separate server hosted website, but the data gets pulled into my blog at build time, so they end up in the same place eventually.

valkey appears to be the largest open source fork of redis that was forked just before their transition to the new source available licenses. One notable thing missing from the readme is how to run with docker, which I saw in the valkey-py docs. You can install the python library with

python bindings for valkey, forked from redis. one notable difference I see from redis is that you can install with libvalkey to autmatically get faster parsing support. For faster performance, install valkey with libvalkey support, this provides a compiled response parser, and for most cases requires zero code changes. By default, if libvalkey >= 2.3.2 is available, valkey-py will attempt to use it for response parsing.

Interesting thoughts here on blog post titles, do we need them? They are so ingrained into everything. It makes me think about markata.dev. I don’t require you to add any meta data to your post, you don’t need a title at all, but you do have to name a markdown file, and this does end up being your title if you don’t set one. Titles are a lot of pressure! I think there is a reason that the big text-based social networking sites (Mastodon, X, Facebook, Threads, LinkedIn, Bluesky, etc.) don’t have titles. Especially for short posts, the title just isn’t necessary. Just say the thing. Interesting observation what rss readers do without one. My own favorite[rss reader], Feedbin, shows the author of the post as the title if it’s missing. Eh, not great not horrible. Hilariously he puts a title on the OG image for the post. I was interested in seeing what would happen in signal, it appears to be showing the author name as well. Confirmed the page has no title or does it? It makes me wonder where Signal gets this title is it from the title block or the site_name meta data

cool article for setting up talos linux with an nvidia gpu. What a wild world it we are living in where these devices that started out being only for hardcore gamers are becoming commonplace in servers and slowly entering the homelab space.

Damn prime makes an interesting point near then end of this video. He’s seen a bunch of people able to just throw down charts and shit at their company and end up being “the coding guy” cause they proompted something once. In a way I can relate, I got into software in a similar way, but at a time that it took a lot more hard work, understanding , and copy past from the right stack overflow. Based on some of the people around me at the time I can only imagine how some people must feel like they got pushed into it without wanting it, and now are building something they don’t know anything about with no care about it or care to build any expertise. Is the future proompted charts from enterprise chatgpt or do we only continue growing more need for software from here.

Tailwind has the best color system, very well done. Even if you don’t use it, it serves as a great color picker.

Big fan of Primes setup. I was not far off of his setup before he really came on the scene, but I’ve picked up a ton of nuggets from him and how he operates. I took his first developer productivity course on Front End Masters as it came out. It is interesting to see him roll back his ansible scripts for bash scripts here. I converted my setup to ansible after watching his first, but have also since rolled back to bash scripts for quite similar reasons. Ansible is great for remote tasks that need to be done on a fleet of machines, but like he says here overkill for this purpose and ends up something that you need to read the docs for every change to your dotfiles. Unlike prime I’ve really leaned harder on installing everything in a docker image and developing out of a docker image. I’ve long built docker images of my dotfiles with the idea that its nice to be able to just use them on other machines, but it rarely happened. In the past year I’ve moved bazzite, an immutable distro. It comes with podman and distrobox, so I install very little on it, a few flatpaks from the store for brave and signal, but most of what I really use day to day comes from my devtainer. It’s nice that I really have one install target for all of my scripts so they become quite stable. I don’t need to worry about arch vs ubuntu vs fedora, no matter where I am its the same base image. I’ve also really started to lean on kubernetes, it is so useful to just be able to start a pod in k8s using the same exact develop setup as I would have locally. Nothing needs installed, I can just bring my dev setup to the cluster where the network and data I might need to debug is.

Jhey has the coolest webdev demos!

Today I learned that cloudflare free tier universal certs do not support multilevel subdomains. By default, Cloudflare Universal SSL certificates only cover your apex domain and one level of subdomain.

Migrating from kedro 0.18.4 to the latest version involves handling the deprecated OmegaConf loader. Switching over does not look as bad as I originally thought. installing kedro 0.18.5+ set the CONFIG_LOADER_CLASS in settings.py swap out import statements config must be yaml or json getting values from config must be done with bracket __getattr__ style not with .get any Exceptions caught from Templated config loader will need to be swapped to OmegaConfig exceptions, similar to #3 templated values must lead with an _ Globals are handled different OmegaConfig does not support jinja2 sytax, but rather a ${variable} syntax

Prime mentioned on stream that Whites were his favorite switch. I tend to like lighter switches and want to give it a try. I really like my Durock lupine’s at 55g, the box whites are 45g, that feels like it would take quite a bit more control, floating over the keys.

A good reference of common screen ratios. I just realized that 16:9 is also 1.78:1. I’ve been putting some images on my blog again, and thinking about using some 2.39:1 ratio on them.

There is a glimmer of hope out there that normal people can scrap together enough gpu to really run the latest models themselves. The ui really appears to be having huge leaps forward such that doing things like rag is no longer such a research project that it was just a few years ago. So excited to see Prime go through this homelab exercise.

I fully believe in our right to repair, ewaste reduction, and bringing a second life to still good hardware that is not up for it’s originally intended purpose. This is a sick console like experience you can strap to the back of a tv, throw in your back to take on a trip, or leave stuffed in your vehicle to game in the backseat. Sucks that it cant do 4k, but I’ve used mine on large screens, and it does quite well for a lot of games, maybe not AAA, but the cartoony multplayer games I play with my kids do quite well.

Damn these deepseek memes go hard. Wild to see openai get played by their own game. It’s crazy that the normie news that I have seen on deepseek shows that the Chinese made what the Americans did at a fraction of the price, without taking notice that they are building on the shoulders of openai.

👏👏👏 This one is really good. I’m right there with him on most of this. I am very hesitant on subscription models, and all the ai tools feel like they are getting ready to be the next round of death by a thousand cuts, this time with pretty limited free tier and relatively high prices to run. I’m sure we will see companies get taken by huge bills soon by building off of someone else’s service. On the flip side I’m definitely the guy that gets in a rut of just copy paste to the ai, wait for codeium to to inject. I feel like I have issues of momentum more than anything. When I’m on one side or the other I tend to stick it out for too long, but less so on going without because that llm drug is calling you when you hit a hard problem. I’m excited to see him build out a homelab for llm stuff that he mentioned at the top. I’m interested, but probably not building one out for myself until we start to see some cheaper maybe used hardware to do it.

Kelsey says several times in this interview, you don’t need kubernetes. If you are running one node you don’t need kubernetes. My question though is, would you use kubernetes? Ya I get it if you are a web developer, data scientist, backend dev, but if you are looking to bee a whole ass engineer, or infrastructure engineer, you know kubernetes, Should you use kubernetes on single node?

I came accross from_attributes today it allows creation of pydantic models from objects such as a sqlalchemy Base Model or while nesting pydantic models. I believe in the past I have ran into some inconsistencies with nesting pydantic models and I’ll bet one had from_attributes set and another did not. Arbitrary class instances¶ (Formerly known as “ORM Mode”/from_orm). Pydantic models can also be created from arbitrary class instances by reading the instance > attributes corresponding to the model field names. One common application of this functionality is integration with object-relational mappings (ORMs). To do this, set the from_attributes config value to True (see the documentation on Configuration for more details). The example here uses SQLAlchemy, but the same approach should work for any ORM.

Dang strong stance that tmux and zellij should not exist. I really do get his point though. Theres a good number of terminal features I often miss out on because I run tmux. Its an app that runs apps, and doesn’t let all of the signals back to the host. But its fantastic at what it does, and brings so much to the table that the little bit of downside it brings is well worth it to me. The other thing missing in this discussion is that I can take my hotkeys and session workflow to any machine just by running tmux. I do not need to run a certain terminal, or install it headlessly on a server to get special features just for it.

Really good listicle of new modern top python libraries from 2024. Very well done article with images, links, and an actually quality listicle with many things I’ve never even heard of.

Good overview of seaborn color palettes. They have all sorts of different types, some designed to purposfully give each color the same weight for catecorization. Some designd to give linear differences in value, some have a parabolic feel with a diverging nature.

I’ve never seen or needed to use a serversideapply in kubernetes before, but I ran into this same issue in my k3s homelab while installing cloudnative-pg. You can do it with argo and you can do it with kubectl

Nerdfont cheatsheet is a fantastic way to copy paste icons into your shell. I just used it to juice up my starship prompt with my current $NVIM_APPNAME managed by nvim-manager

For my next drive upgrade in my homelab I am gong to be using one of these factory recertified drives from serverpartdeals.com. Found them on an LTT video awhile back. They are some lightly used and recertified, fully burnt in drives. Shop for drives that are certified once again by the manufacturer to work like new. Factory ReCertified drives are cost-effective alternatives compared to factory-sealed new counter parts. Additionally, unlike in mass production, the re-certification process involves closer attention to the overall operation of the hardware so that the re-certification will not have to happen a 2nd time

dust is one of my favorite rust rewrite tools. Its so useful for narrowing down file system bloat and cleaning up some disk space on your nearly full disks. It runs right in your terminal and gives you a nice bar graph on the top directories in use.

Keycloak looks like an interesting way to setup sso. It’s part of the cncf so it’s got a good backing. I want something better for argo workflows and this might be it. I’m curious what else I can tie into it.

Don’t stop learning! Stop trying because you have a doomer outlook on ai, llms, industry and think they are taking over. If you have no hope for the future, if you stop now you are cementing in that you will be no good and the ai will be better. Many, maybe most of us in this industry go here by hard work, long nights of learning, trying to solve problems that our job had. If llms take over then the world is going to be a whole lot different, it will be a world you cannot predict or plan for. For now put your head down and succeed in the world we have today. TEEJ has some great thoughts on this whole sentiment, put this on for you morning walk or whatever you do.

I like the charts that Theo brings to to these videos. Shout out for a positive k8s reference and not shitting on it. Htmx brings html/css just a bit further down the complexity graph with little to no extra effort, while react allows us to go all the way full complexity at the cost of build and dev complexity to go from zero to 100 as soon as its introduced. htmx brings us back to the ease of jquery ajax without any complex swapping or json parsing, all of the object parsing and html templating is done in the backend, the front end just tracks where to put it. HTMX couples the frontend and backend much tigher, since all of the front end html is generated in the backend, done correctly it is not possible for the front end to get out of sync and try to do things that the back end does not know how to handle, vice versa.

It’s interesting how many people in tech maintain a blog. I think part of this brings us back to web 1.0 days when so many individual websites owned the web it was a free for all unindexed land and you got to own a small piece of it. I agree with most of Brittany’s points here I write a lot to keep my skills sharp, and to refer back to. Brittany mentions keeping all her old posts, even the cringy ones. I’m all with you here, I’m just wodering how you look back at anything you wrote in the past and not get a bit of that feel, maybe its just me, but I see cringe and mistakes gallore, but it all makes me better moving forward.

nice overview of availability measurements and what they really mean. The crazy world we live in today depends on so many things runnig, its also so hard to measure your uptime, The uptime metrics can mean a lot of different things. The site is up and accepting traffic, but can users make changes or submit orders, there is a lot more to it than just up or down. I really appreciate Brittany’s story from Nike nested in there.

The SuperMini nrf52840 is a sick controller for building keyboards, affordable, easy to get, and compact. Bluetooth and wired setup just works in zmk. This page has a nice image of the pinout.

Bluesky is almost excatly like twitter was when I joined years ago. It’s gone crazy lately bogged deep in politics, bots, and ads. I’ve seen like two scroll pages of ads in a row, its nuts. What I did not know before Joel pointed out here is that the feed I am looking at is my following feed, its only feed of people I follow in descending order. On bluesky you get to pick your feed!!! This feels like tweetdeck did back when we were able to run that. You could tune in search terms and save them it was glorious. Bluesky has some really interesting ones that you can use like popular with friends, only posts, my bangers, that have a pre defined algorithms.

This man is responsible for making gaming on linux what it is today. Such a heartfelt story to hear that reviving his dad’s machine was at the core of what drove him to do what he has done for the wider gaming on linux community. Update on your schedule, remove all the tracking and bloatware, this is what drove him to start using linux before making it accessible for his Dad. But really do update, this is not your 2002 PHP box, things need updated and regular updates help the process.

Wild that the podman-compose github readme calls out k3s as an alternative. compose definitely has its place, especially for local development on a developers machine, its so much easier to stand up and get things like hot reload up and running smooth.

Css is getting so good, new things like interpolate-size are making things that use to require some deep expertise and hacks intuitive and easy.

Looks like a great start to a rules file for fastapi. Place the happy path last in the function for improved readability. I have never heard anyone say this. It feels weird to me. The other early return, find and handle errors early all make sense to me, but happy path last is new to me.

Really interesting way to generate a rules file for agentic workflows based on your current repo. John uses gitingest here, looks like a fantastic tool, but probably not useful for most private repos. I’m sure you can replicate the same thing in a private repo wtih a small amount of effort the few times you need to do it. gitingest looks like a great way to pull in some extra context for some open source dependencies that you have though.

Gitingest has a python package on pypi that you can run with uvx, and it accepts the same arguments as the web version, right in your terminal

Replace hub with ingest in any github and get a prompt friendly codebase ready to feed into any llm. It combines the entire codebase, based on a gitignore style glob that you pass in, into a single TXT file.

Definitely need to give codecompanion.nvim a try, it looks like a competitor to windsurf but in nvim. It looks so feature complete that its hard to grasp all of what it does.

Great panel of software folks at open sauce. It was interesting hearing from all these creators from the perspective of an open sauce panel.

New release out for nvim-manager that supports installing pre-configured distros. It’s such a breeze to install these now, its been fun to go through each of them. The currently included distros are. LazyVim AstroVim kickstart NvChad LunarVim

First release of nvim-manager is out. Your dotfiles change a lot, sometimes it’s hard to manage all of the places you have installed them and potentially made hand edits to. nvim-manager allows you to easily make static releases of your dotfiles, and keep your nvim install from breaking by leveraging NVIM_APPNAME and pinned releases of your dotfiles stored in ~/.config. In this directory you might have many nvim configurations installed, nvim-manager automates the process of installing and updating from your dotfiles, while keeping previous pinned versions untouched.

Nice list of url escape codes. I did not actually know that to get a literal sequence like %2D you can use $2D.

This setup fixed my nvim syntax highlighting in helm templates.

imagegenius has made an immich all in one setup that looks much easier to use than immich.

I can say I had the same kind of feelings when I first saw something called “Own Your Web” being run in Buttondown. I totally get it. It takes time and effort to build your own stuff, email sending is hard, not done right ends you in the spam folder. There is something about the name though that I think needs to set an example and self host as much as it possibly can. The changelog has covered this several times, do they need to go to the crazy lengths they do to run their site, no probably not, but it keeps them in the loop. They are using the tech they talk about in a very real and production critical way to run the show. Cant wait to see more from ownyourweb.site

Gyroscope better than a mouse?? Nerd nest really sells how having two gyros in the way they have done for noise cancelling changes the game on it, and makes it a contender to replace a mouse. It really makes me want to try it. I love how repairable this controller looks. I’ve got to imagine that the fact that it comes as a kit, and all the parts are available that this hits S tier repairability. My current controller of choice is a PS5 and I’ve had stick issues I wish I could fix. No analog triggers, I’m out. Maybe they will make it an option in the future idk. I don’t play shooters where I need a hair trigger, this won’t work on session or driving games. micro usb, seriously, that kinda kills it for me too. edit I looked on their website and they have usb c in the latest version

postiz looks like a very polished way to automate and schedule posts to all the social services.

Reminder to myself, look into self hosting postiz with this helm chart later.

Chris Coyier had a small re-align on his site, some good nuggets in here. I like the idea of having a photo of myself prominently on the site, so you know who you’re dealing with here. I really like this after thinking about it and I think I am going to make sure I get my face back on my posts. I do have my 8bit style pixel art image of me that I use on social media, but no real picture. I feel like a lot of people redesign their entire website when it’s time to update to the latest list of social networks and I’m no different. Once you touch it you gotta keep going. I can totally relate to this, once you open the thing, you get the build tools greased up, and your confidence high that re-deploying isn’t going to mess something up, I tend to start digging in to other things.

Very interesting way to catalog games, I need to make a catalog of mine, I’ll probably start adding some blog entries for games I’m in and have completed. Wonder if there is a way to hook into steam with python to get achievements and progress live.

Tons of cool people came out with their rss feeds here, again will need to browse more closely later.

Sturobson has a ton of rss feeds here, I recognize quite a few, will definitely need to poke at some of these later.

I’m a sucker for good own your own shit on the web blogs, and Matthias Ott has a top notch one here. The archive has been a great read so far, I’ve discovered things like slashpages.net.

What a great first guest for FAFO. Kelsey is always such a great listen. He talks about killedbygoogle and how engineers get no incentive to work on old projects, google had no incentive, and you got the thing for FREE in the first place. He talks about end of career and having love you money, having so much in the bank you can say no. If you are presented with a project that does not align with your values you can say no.M Justin even mentions how Google has more killedbygoogle projects than Amazon has total projects. If we knew how hard it would be, we would never build it. Autumn’s Fav quote from ep1 of shipit

Well done changelog with some really good inspirational nuggets. Many slash pages I want to check out and an 11ty contribution graph.

A nice list of slashpages you might want to consider including / aliasing / 301ing. These feel like nice things to setup and keep in the back pocket for obsidian style wiki link to easily. I get kinda bad at wiki-linking as much as I would like to, mostly because it does require some amount of work to make the page, and keep it up to date over time, then remember that you even have it. Some are serious, some very common, some quite useful.

Theo does a fantastic history of serverless here. Theo can’t have an infra video without shitting on k8s. Specifically people who have never touched k8s pushing fear of k8s to large audiences of people who have never touched k8s. If you are a webdev who solely lives in webdev space and never touches as much as a dockerfile listen to him. If you touch infra at all try it before you take his opinion at face value. If you plan on having traffic spikes 10x your regular traffic for something like black friday, serverless might be right for your use case. He argues that targeting a stateless deployment of serverless leads to better code. I’d like to see more examples here. Maybe most of the code bases I work on already do this. I’ve never targeted a serverless deployment, but I’ve targeted horizontally scaled deployments many times and they feel like they have the same targets. For instance if I spin up 8 pods for my application or uvicorn with 3 workers I have to target statelessness, all of the state must live in the database and cannot live in memory. Even if I target 1 instance in a containerized environment I have to be ready for restarts at any point in time. I might be missing something here, but I don’t see how this point applies to serverless. Companies that can run on scale to zero can allow your shitty side projects that have no users run for free indefinitely because it costs them nothing. He compares planetscale running mysql vs Turso running SQLite stored in s3. He compares early deployments of LAMP and MEAN stack running on one server then jumps to serverless. It feels like he is missing the angle of owning your own hardware. This might just be the experience difference between theo and DHH. DHH works on one focused company, Theo is jumping around between startups. At the end he covers a dumb side project that has branch deploys. It had a 100s of deploy still running, some very old, some never even touched, vercel can do this because it costs them nothing if its not running. What I didn’t think about right away is that there is probably a backend component to this. In my python backends I often have deploy environments, but since they don’t scale to zero they have at least some cost even if its small, with this I cannot just keep hundreds of them running because that cost would add up. I would argue that this is hardly useful, because you can checkout any old version and run it locally, and you rarely are going to need to poke through these old branch deploys. But in that case that you need to quickly do this serverless is definitely superior.

cool video on expanding vim with cli.

Lane from boot.dev madde this fantastic video about serving files on the internet. It has me wondering if I need to rethink a few of my things that I have built. I have a few things I am serving media from, but I have very aggressive cloudflare cache rules on them, so each file should only be uploaded about once per year. My problem going straight out of minio right now is how do i set headers for cache control on it. If I can’t set the cache control and everything is coming out of minio this does not solve my problems. I went back and played with presigned urls and you can in fact control and set response headers, this is definitely the way and I have been wrong.

This is a cool new feature coming to bash, I can’t think of a use case I have out of the gate, but it looks cool. I’m thinking this might be good to keep in the back pocket for something like CI where I don’t have a hightly tuned bashrc file, and I want a dynamic curl request based on some state that exists as a fille.

Nicely worded Cassidy! Javascript can be too much, it can bog down low powered devices, we can ship so much that its untenable on poor connections. It can be argued that its a bad language and putting it on the server is a mistake, but it brings sites to life. It makes it possible to extend the static nature of html/css with just a little special spice only your site needs. I’d add to the argument that a lot of js should go away over time. Over time libraries such as jquery have fallen out of use, not because they are bad, or have been replaced by new libraries, but because the browser has adopted most of the functionality that jquery brought. As a primarily python dev I’d really like to see htmx die a very similar graceful death. There have been several iterations on this idea, and the crux of it is very similar, give the ability to use HTTP verbs right inside html with some instruction of when to apply them and what to apply them to. No js should not go away, it never will, we will always find new patterns that the browser should take, sites will always need a small bit of spice to bring them to life.

/p warp will warp your party to your current server in hypixel skyblock

WTF, you can just change a server response from devtools and update a vuejs app? Just tried with htmx, and my GET requests are not showing up in the sources tab. I’ll keep this in my back pocket for next time I’m supporting a vuejs app though.

Gary Vee, is the goat. leaving this image as a reminder for myself

Atuin v18.4 is out with an atuin wrapped command. Here’s Ellie’s Wrapped. This is a pretty sick command, I’ve only been using Atuin about as long as I’ve been on bazzite, so maybe 4 months, but here is my wrapped. Comparing to Ellie, I’m surprised that I even have 126 package related commands, being on an immutable distro most of my packages come from the container, I don’t need to run packaging commands. My cd is also super low, I use tmux sessions to take me where I need to be most of the time, which is a project’s root directory, It’s pretty rare to see me outside of a project root directory. I’d venture to guess that nvim would be higher if it didn’t auto start on every tmux session as well.

Red Hat has donated the whole open alternative to docker to the CNCF, the hosts weigh in with thier opinions all being pretty positive as they seem to be a legitimate donation and not dumping crap on open source. podman podman-desktop buildah compose-fs podman bootc scopio Personally I’ve been running podman exclusively at home since switching to Bazzite in August. I’d like to use buildah in ci. I gave it a try, but was unable to swap out my use of kaniko for buildah, I will get there, but it was not quite as drop in as I wanted.

Holy shit these AGI models are incredibly expensive to run, require lots of wild hardware that there is not enough to go around, and requires shit tons of power to run. Now more than ever is time to distinguish yourself with deep expertise, jack of all trades is being eaten by ai. People with deep expertise are getting a jack of all trades bump from ai, not o3, just the regular stuff.

Theo weighing in on the 2025 job market. It’s no 2018 out there right now, the ratio of jobs to engineers in the market has flipped big time. Theo as usual really focuses on community, being in a community, and being a good citizen. At the end of the video Theo weighs in on his experience hiring, and generally it starts with we need someone to do x does anyone know someone, then goes to a more formal internal post, then more formally asking internally does anyone know someone, then maybe to his community, and if he really still needs the person it might become an external post. These days there are so many good engineers on the market that very few good jobs actually get a posting for in his opinion.

kitze hating on python packaging in new ways. Python packaging has a lot of quirks that can make it infuriating. Not once have I thought “you know what this needs, quotes and braces”

This tip of using tinkercad to do boolean operations on an stl of a solid gridfinity bin and an outline is absolute fire 🔥🔥🔥. This feels like a relatively simple operation, but to do it to a generated stl proves hard to do in most modeling software, at least harder than it needs to be. Somehow tinkercad got it right and made it a very basic operation to do. It took me a minute to find the Merge button that Uncle Jessy mentioned, they call it a group in TinkerCAD.

Damn Glorious Eggrolls is still making gaming on linux better. Of course its containerization that drives everything on linux these days. This is a pretty badass talk. Umu is already running in steam and bazzite. Bazzite gamemode uses this to get a gamescope session running.

Great list of 4 tips for running fastapi applications. Fat routers with all of the logic built in makes them hard to test, hard to refactor, causes lots of duplication, and makes it hard to reuse the business logic code later in something like a cli application. I really like this advice! He reccommends deploying as early as you can get a healthcheck live in your application. I’ve found too many times developers build something that is really hard, or impossible to deploy, when if they had tried to deploy early they would have spotted some easy to fix issues. This is less important if you are building out of a template that your team commonly deploys from, but very important with new patterns.

This is really interesting, the lazy uv scripts are really becoming quite appealing, especially for something like this to just pop out of an llm ready to run. The article features several examples of these one-shot prompt ideas that I suggest you give a try, and a prompt for creating them.

Here the integration to depot appears to be opt in using the --depot flag on fly deploy. This must have changed over time though because today it was giving me issues and I had to opt out using fly deploy --depot='false'. Looks like a great service and I just learned about them on their bad day.

Just learned about depot today ironically because it seems to be down and fly is using them under the hood to do the container builds, seems like a really great service for fast builds accross your team.

This was an eye opening video into agentic editing workflows. Dfferent ai tools use different rules files, windsurf uses .windsurfrules. Test out your rules file by having it say something at the beginning of the output to verify that the rules are being applied correctly. He suggests to use this key rule for debugging purposes, otherwise you are guessing to what rules if any it is following.

This is a pretty great episode talking shop with typecraft. They talk setups, cameras, content creation. I found them talking about their linux setups particularly interesting. Dax talked about his flow from building his own machines to using reliablesite.com. Dax hates on c-a, both typecraft and dax use c-s, which normally freezes a terminal, we can all agree that is useless. I use the default c-b, it seems fine for me. Ubuntu Ghosty Tmux Nvim Ruby Dax talks about terminal.shop and how they originally planned to have a web front end, but after they had so much success they stuck with it. now they are leaning harder into it and are building out integrations with a bunch of languages and an api, but no front end.

PEP 723 is what is inspiring all of these lazy self installing python scripts, Authored by the author of hatch and pyapp. This is a really cool thing that uv has picked up and made python packaging just a bit easier.. maybe… dependency resolution still sucks.

I’ve kinda fallen out of using direnv now that a lot of my projects use hatch, I generally just hatch shell into them. I just need to make sure I go through all of them and make my installer uv. Now I’ve been thinking about making uv my only needed dependency to run a python project and leaning more to something like uv run --with . uvicorn myapp --reload

I really like Trey’s steps to making an executable python script with uv his old process seems to be the same with a new shebang Add an appropriate shebang line above the first line in the file (e.g. #!/usr/bin/env python3) Aet an executable bit on the file (chmod a+x my_script.py) Place the script in a directory that’s in my shell’s PATH variable (e.g. cp my_script.py ~/bin/my_script) And here is the new format the the shebang followed by the metadata comment block defined in PEP 723.

I still cannot believe I thought I had such a unique idea with thoughts only to find out shortly thereafter that Simon beat me to the punch by about 20 years! Simon put a lot of work into this post please check it out. Its not only filled with tons of little nuggets about blogging, it has just a ton of links to other posts I’m itching to read now. Simon mentions it in the post as well, but one thing about having a link blog does is that it proves that you actually read articles that you share. This is really more about proving to myself, and reminding myself that anyone else, it’s a good feeling. include names of people, and add tags for common people something extra - the whole point is to bring something to the post without stealing thunder, but rather linkbuilding for the author that put the hard work in. copious amounts of quotes It should bring the author up not put them down well .98 posts per day Simon mentions starting in Nov 2003, which puts that at 7722 days since then. That is so close to a post a day for 21 years, even though they are often as short as one sentence that is impressive to keep up over time. So far I’ve published 7,607 link blog posts and counting. I just check and I am at 513 days in with 413 posts (0.8 posts/day). Not quite up to Simon, but close enough to be proud of. My only grief is that I didn’t start sooner. I have a full post explaining more about how I do mine and the stack behind it here. This is one of the main reasons that I built dropper for myself, and plan to include it in fokais as a product to help others with their blog. This is definite evidence that I am not the only one with this issue! I’ve tried half a dozen or so easy techniques from obsidian to various github repos and none have really hit me with inspiration to really add screenshots like I do with dropper. I write all of my thoughts posts in a chrome extension drop down local first style and when i hit paste it uploads and gives me a markdown image link, it cannot be easier for me. When I store them I compress them and convert them to webp. I serve them by streaming them out of an s3 bucket with fastapi and some very generous cloudflare cache headers, so I should only get a request for that image about once a year. Warning, I built dropper for me, no guarantees that it works forever, foakis will be for more users. I am loving bluesky more and more, learning this just gave it another level up. I just noticed that Simon has Series on his blog in a very elegant way. One I need to get back to blogging more, I did not do so good in 2024, and two this is awesome and I want it. test