Ricardo Martins

Ricardo Martins

@ricardomartins.dev

someone who dislikes writing about themself in the third person. does computer nonsense for a living. occasionally locked into a meat suit. πŸ“ Algarve, Portugal

18 followers36 following35 posts

> Typing is a sign of bad architecture, bad documentation, or/and lack of ownership. that's… a take, for sure. why do some people get so incredibly mad at optional types in Ruby? like, it's fine to say they don't see the value but to make accusations like this? wild.

Ricardo Martins reposted

Making myself angry for no reason at all remembering the American transit planner that went to Amsterdam and was appalled at how little stakeholder engagement they did in favor of actually building shit.

Ricardo Martins reposted

Replying to @hikikomorphism.bsky.social

2026 is wild because you can just say shit like this and everyone will be like "oh yeah, the text elementals, that's a good phrase for what they are" instead of looking at you like jesse what the fuck

right, count on me to forget to mention the key detail: apparently some devices just… default to using the best encryption level that’s being advertised in the ssid (which is good) regardless of the actual bssid they’re targeting (which is bad)

Ricardo Martins@ricardomartins.dev

today’s cursed knowledge: don’t mix access points advertising the same ssid with different subsets of WPA

my sister had one ap advertising wpa2 and wpa3, and another one advertising just wpa2 all devices worked fine around the first ap but laptops and cameras had weird connectivity issues around the second one the only thing that fixed it was ensuring every ap was configured to wpa2 only

a very special β€œget bent” to microsoft for making the wifi network show up as full signal strength in the taskbar (because the laptop was close to the wpa2-only ap) but then showing minimal strength after connecting (to the other, very remote ap)

Ricardo Martins reposted

Replying to @shimminykricket.blacksky.app

Basically AI allows software engineers to produce shitty code *faster* and in higher volumes which increases the amount time spent on code reviews and revisions nullifying the productivity gained from producing the code in the first place.

just created an account in hugging face and I noticed something in the profile section… "Set your P(doom)", eh? really? y'all need to go out, touch grass and stop letting the text generator have access to important stuff. stop shirking your responsibilities as technologists.

Ricardo Martins reposted

"A.I. is making me a better doctor. And I worry that it’s making doctors-in-training worse."

nytimes.comOpinion | Being a Doctor Will Never Be the Same After A.I.I’m confident in my mastery over what doctors do β€” and less confident about my continued mastery of an ever-evolving universe of facts.
conputer dipshit@davidcrespo.bsky.social

it feels like every field β€” engineering, natural science, social science, humanities β€” has basically the same problems around AI, with the same solutions, yet everyone talks like their problems are due to the unique character of their field

Ricardo Martins reposted

This is an unpopular take but as a neurodivergent self-employed person, LLM's have been an unbelievably helpful tool for the things I have struggled with my whole adult life: managing my finances, filling out paperwork, and staying on top of my email. I'm functional in a way I never dreamed of.

despite/because of the warnings against it, I watched the rails world keynote and uh… that will take me a while to unpack totally get why people would want to purge their memories of having watched it

dude spent so long boasting about rewriting all software he missed into something tailor-made to his needs and taste, yet he missed that is an argument for the death of all saas except those related to serving llms "look guys, software is now worthless, so my software company has a bright fut… oh."

Ugh, hearing Nick Bostrom talk about LLMs as if they have different levels of "cognitive sophistication" just doesn't sit well with me. Stop anthropomorphising programs this way, damn it.

Don't get me wrong, I recognise that LLMs are useful tools. It would be absurd to refuse to do so at this point. But they're text generator programs, you can't forget about that. They're not "minds", they don't have "cognition". At best they're like the mindless aliens in Blindsight.

Ricardo Martins reposted

Replying to @mskellymhayes.bsky.social

The most generous assessment is that this was shitty, lazy engineering; the more in-character scenario is that the sandbox was intentionally breakable because of the story it would generate. Whichever you believe, the story is not "OMG, they made AGI!"

Ricardo Martins reposted

The Anthropic, Meta, and Google "breaking containment" incidents all resulted from setup errors (or a stunt) by the same company: Irregular. Irregular created a sandbox that wasn't really a sandbox because it had a live internet connection, and gave instructions that were then carried out online.

homebrew dropped support for x86-64 macOS to tier 3, so I guess it's finally time to consider buying a new laptop 🫀 brew.sh/2026/09/13/h... on the one hand, I've been grumbling about this laptop for a couple of years. on the other hand, it still works fairly well

brew.sh7.0.0Today, I’m proud to announce Homebrew 7.0.0. The most significant changes since 6.0.0 are faster installations and upgrades, stronger sandboxing, a native macOS app, built-in vulnerability checks and ...

Having fun (read: torturing myself) discussing with Astra how to host a PDS on my home server with transparent failover *and* failback, minimising manual maintenance (i.e. how to recover from a failover situation with minimum toil) for the least added cost. The model's surprisingly good at this.

Kinda converged towards: - use tranquil instead of reference PDS - edge ingress on Fly - route request primarily to home server via Wireguard - route to tranquil-on-fly as fallback (auto-scale when primary fails) - Backblaze B2 for object storage - neon.tech for PostgreSQL and reliability

The problem is the database. Neon's free plan doesn't cover PG running 24/7. If I keep the DB only in my home server, I can't failover in the scenarios I'm considering. So I guess I'll have to either 1. forget the idea, 2. accept downtime if I do self-host, or 3. accept I'll have to pay for the DB.