continuing my journey with NotebookLM, it's so beautiful, but 😂 😡😭
I ask google slides to “beautify this slide”. It generates a futuristic image, fine. but I want to add links, so I ask it how to do it.
it literally generates an image explaining how (completely irrelevant)!
it’s 2026, AI writes code for us. yet somehow I still have to manually align text boxes in slides.
I’m really looking forward to speaking at #Scale23x in Pasadena 💛
I’ll share what we’ve been building in OTel Weaver: you define your telemetry schema, and we give you tooling to generate docs and code, and validate real telemetry against it.
Saturday, March 5 · 5pm, come say hi! 🙂
And now computers can finally spit it back to you, despite it being decades old
Nobody praises you like AI. It’s totally cringe, but hey, useful for perf reviews.
In Nov 2025, we delivered native #OpenTelemetry support for .NET 10.
In this post, we share a behind-the-scenes look at this journey, and a few updates about how we're helping those who use OTel signals within #dotNET applications move to .NET 10.
Rust makes me deeply grateful for everything other languages don’t force me to think about
you've used to have this cool background with a couple of Jesses behind the 'real' Jesse. Copilot is watching!
feels very productive though, 10x developer
Exponential are a better default, otel uses explicit ones because of back-compat. All backends should expect to receive standard semconv metrics with explicit or exponential boundaries, so you should be fine emitting exponential by default. They are still semconv compliant
Maybe too ambitious 😅
Exponential histograms don't need boundaries and all uses can opt into exponential histograms. By default Otel uses explicit boundaries (back compat) and instrumentations provide default boundaries. Users can override or pick exponential
Faas conventions didn't go through clean up yet, so unfortunately are inconsistent
That's the reason. We try not to duplicate, in theory we could report count too, but that's be opt in and not free for user. Cheaper to learn once than pay forever
I think I know why AI generates walls of redundant text.
It doesn’t know humans have long-term memory, so it does aggressive context engineering on us.
Join us for our first joint #Grafana ❤️ #OTel and Tempo Community calls on practical distributed tracing from the field. Bring any questions you may have!
See the YouTube description for more details.
Thursday February 12th, 9:30a PT/6:30p CET
I ask AI to fix my grammar. It adds articles.
Next pass, it removes the same articles.
How exactly are non-native speakers supposed to learn English?
If you are at FOSDEM, let's catch up!
Productivity hack: put a few things on today’s to-do list that you really don’t want to do. Procrastinate the worst one by doing the others. Works especially well when the world feels like it’s falling apart.
We should move all tech conferences to Canada
Yep, same here
We ran from Putin’s Russia to escape this kind of lawlessness and abuse of power, but ICE reminds me of it in unsettling ways. It’s scary. It’s not going to fix itself. No one is coming to fix it for us. And I still don’t know what I can do that would actually help change things.
I love GenAI observability because it’s a greenfield - no one really knows how to do it right yet. And I had a lot of fun learning how we observe Grafana Assistant!
A mentee got caught up in a layoff. 😓Who is hiring software engineers in Canada? Focus has been in the o11y / OpenTelemetry space, most proficient in golang, but they’re open to new domains
Whenever I need to check how to do things idiomatically across languages, I look at the Azure SDK repos ❤️🐬
Feels a lot like “everyone who wrote this has already left”. Someone hands you 3M lines of code. Maybe there are tests. Maybe docs. Maybe CI. No backlog. No context. Nobody knows what works.
Yes, but some events are special - they have unique context and duration and we call them spans and propagate their context wherever we can. Context should be the main pillar actually, not traces
Writing walls of text has become way too easy lately. Reading them still costs just as much
Found this note from my 5-year-old. I did something bad. No one remembers what, but I broke his heart, officially documented (if you can decipher)
