Wanted to look H Dreyfus and realised just now he’s the sweet voice from the Philosophy 185 podcasts I listened to some 15 years ago. So happy that this is now relevant for my day job!

Finally realised MCP tools don’t feel right because they are not ready-at-hand for the model. Time to move from Frege to Heidegger and Wittgenstein in NLP

Omg yes! Linear digressions is relaunching! Never tidying up the pod feed is paying off..

I have a todo to setup a fallback static feed for when my feed isn’t up, clearly over engineering considering the discovery feed is fine without a fallback 😅

Decided to go VPS instead of homelab, ovh looks suspiciously cheap for the 24GB VPS-3.. is there a catch?

The boilerplate for tangled.org/@aleeve.dev/... finally works! Now I just need the last few steps for the query in the continuously populated postgres.. next weekend

@aleeve.dev/ott

Scalable and distributed custom feed generator, ott - on that topic

tangled.org

I wonder how many feeds actually check audience of the incoming request getfeedskeleton request? If most don’t a nice and simple custom feed would could be a proxy that filters or mixes other feeds. Could be a simple cloudlfare worker

How are people running their custom feeds? I went with cloudflare tunnel directly into my kind k8s cluster, but feel I need to add cilium so the tunnel pod is only allowed to talk to the main xrpc service and I can lock stuff down. Still feel a bit uncomfortable with this..

Jumped the gun a bit here! MVP like80% done..? (lower if counting polish)

tangled.org trending@stitch.selfhosted.social · last yr.

aleeve.dev/ott Scalable and distributed custom feed generator, ott - on that topic ⭐️ 2

Me trying to apply backpressure while consuming the bsky firehose:

Making good progress with the ott-custom feed, but really need to hook in to the bsky moderation.. sooo much explicit content on here 😳

An OTT service, but it’s actually a on-that-topic custom feed

cnpg’s image tag requirements really doesn’t feel that cloud native

Working on a custom feed and eyeing the firehose quite a bit. Thought I was in a bubble, but ”en” posts are extremely dominated by us pol 🙃

When I do my first custom feed I will make sure post from the same account appear chronologically, having the year progress post come flipped is trippy

Am I doing something wrong or are arroyos async udfs quite flaky still? Getting weird casting panics 😭

The Discover feed is full of some Italian bashing the city planning in Stockholm, amazing algorithmic work clocking that this is what I’m here for!

Watching Jon Gjengsets decrusted series and just in total awe! Amazingly clear and mindful, wow!

’Monkey typing Shakespeare’-custom feed, does it exist already?

So today was autogens turn. Long time since I saw a rest client this convoluted.. these frameworks appear to be radicalising each other

Reading smolagents for a work thing. The manual parsing of ASTs for generated code is cute, but wow does it feel like a missed opportunity to showcase WASM isolation

Apparently it’s easier to learn typescript properly and use wasm-bingen to do web-api interactions than using web-sys or gloo.. All I wanted was to try out my shiny new hammer, not this language upsell ladder

My grandad collected autographs at the 1955 Gothenburg Interzonal. Among the players where big names such as Spassky, Petrosian, Najdorf etc. I think Spassky signed on a6.. but I would really like to make out the rest as well #chess

My team won the local tech scene pub quiz! A lifetime of geekdom finally starting to pay of!

It’s never the framework, and if it is it will never be fixed. Except with leptos, because they somehow already fixed it on main always 🤯❤️

Done a lot of coding during the holidays, but as I haven’t actually committed anything I think it technically counts as going offline

Upping the audio project from leptos 0.5 to leptos 0.7. As someone only interested in CSR and using websockets, rtc connections, and MediaStreams in signals it feels like there is an awful lot local colouring.. I must be doing something wrong

Picking up my webrtc live audio jam project. I hope I can trim the latency enough.. What would be the next step if it’s to laggy?

While the zero config concept is cool, I would much prefer it if the ghostty config came with all options present but commented.

For oauth scopes I would really like nsid level read, write, execute. Is that the plan? #atproto