Tom Elliott

Tom Elliott

@telliott.me

DevOps enthusiast. Posts bi-weekly on https://thefridaydeploy.substack.com/, occasional public speaker.

481 followers1.4K following1K posts

Feel like speaking in terms of memes harms your credibility? Upgrade to using "principles" named after people you've never heard of! Like a transitive Dunning-Kruger effect.

Saw Project Hail Mary last night. A great movie and a mostly faithful adaptation. Except they cut out the scene at the start where he rips out his catheter. Cowards.

Just imported a resource into a Pulumi project for the first time. Pleasantly surprised by how smooth it went. I think I'm now a fan.

Historians have recently uncovered a centuries-old artwork depicting Norman soldiers fighting alligators in a marsh. They call it the Bayou Tapestry.

Anyone else think “Commuter Benefit” would be a good name for a band? “Hi, we’re Commuter Benefit and this is our cover of Fast Car”

"Make no mistakes." is a confusing kind of prompt. Confusing in the sense that I can't decide if it's more like: - Giving an intern no coaching and being frustrated they didn't "meet expectations". - "I hereby declare that Facebook cannot use my content for stuff."

CI/CD has been top of mind for me for at least the last couple of years. Now coding assistants are starting to be adopted widely, I'm wondering how the SDLC might change to keep up. Will we be using centralized CI systems 5 years from now? open.substack.com/pub/thefrida...

open.substack.comWe won't need CI in 5 yearsThe way we build software is changing, and the way we deploy is going to have to change right along with it. What we consider to be "good" CI today may be very different from tomorrow's definition

My Apple Watch told me it was 49 degrees today. Lies. I should have looked at my phone before I went out in 17 degree weather without a hat.

Iain M Banks's Culture series totals around 1.4 million words. At 750 words a day, that would take 5 years. They were released over a period of 25. Planning and editing are a lot more work than I thought!

With all this generated code, PRs feel like the bottleneck. Then again, thinking back to my teams before AI, they already were! A PR written in 1 day took 3 or more to approve. So we added linters, shifted left, all that good stuff. We need all that more than ever now.

Wait, you’re telling me that that Chinese robot they’ve been showing off is called the T800? Holy Torment Nexus, Batman.

I write fairly regularly, but always feel like I don't read as much as I should. Looking back over the past week, I've read a bunch of articles, so why do I still feel like I'm behind the curve? Subscriptions. (thread follows)

View full thread

I'll make some effort to read content from friends, you know who you are 😅 Even if I see a really interesting title, when I see the message I'm in correspondence mode, not blog reading mode.

All these subscriptions are making reading a chore. Time for me to dust off the old RSS reader and try to enjoy reading again.

I write fairly regularly, but always feel like I don't read as much as I should. Looking back over the past week, I've read a bunch of articles, so why do I still feel like I'm behind the curve? Subscriptions. (thread follows)

View full thread

Nowadays, I'm also subscribed to a ton of mailing lists via Substack. I'm assailed with emails and (until I turned them off) notifications every day. I rarely click through.

Every day, a pile of little obligations arrive in my inbox, and quickly get buried by everything else. So I never get round to them, and I feel guilty.

I write fairly regularly, but always feel like I don't read as much as I should. Looking back over the past week, I've read a bunch of articles, so why do I still feel like I'm behind the curve? Subscriptions. (thread follows)

Most of what I read I discover kind of organically, via search, social media or the like. Got a few minutes to spare? See what's on Hackernews. There was also a time when I had an RSS reader constantly loaded in a tab and I in idle moments I'd catch up on some favorites. I miss those days.

Started on a just-for-fun project yesterday. Rendering all the buildings in NYC in 3D. opendata.cityofnewyork.us has a ton of great data sets, including this building data! Learned a ton about optimizing scenes. Had to get creative with reusing and grouping meshes. #buildinpublic

Tom Elliott reposted

This take is a bit outdated now. At its core it is still true, but lots has happened to improve LLMs as fact-retrieval tools, and I think they’re generally very good at it provided you, the operator, have your head screwed on and do the work to verify.

Barry Deutsch@barrydeutsch.bsky.social

The only thing ChatGPT ever does.

If you're a typical Git user, you may not have heard much about Git trees. They're a key part of plumbing, but don't often show up to the end user. So here's my overview. Don't worry, this particular tree won't leave pine needles all over your carpet! open.substack.com/pub/thefrida...

open.substack.comLooking below the surface of Git - what the heck is a tree?You may or may not have heard about trees in Git, but you've definitely worked with them! This post provides an overview of what trees are, how they fit together and walks through an example.