rsanheim

53 ms·

1.
by rsanheim 20d ago
"seam" dates back to refactoring & software engineering literature that pre-date LLM use by 20 years -- see Working Effectively With Legacy Code for one example ( https://www.oreilly.com/library/view/
2.
by rsanheim 1mo ago
AFAIK there is nothing like this out there, because there is not one standard for a "pull reauest" in the sense of how a web based forge does it. and it gets complicated quickly due to complexity of git and forks. the closest thi
3.
by rsanheim 4mo ago
filter is also just filter in ruby - aliased as select (and find_all) on Enumerable.
4.
by rsanheim 5mo ago
> has Gas Town shipping something in public that can be evaluated without all of this surrounding hype and blogposting? No.
5.
by rsanheim 5mo ago
This post sets up a straw man from the outset, and only gets worse from there. I understand how we got here, where many experienced programmers, managers, and bloggers only know capital-A Agile as the watered down version sold via certifica
6.
by rsanheim 5mo ago
Bragging about loc was silly before ai. Now it’s not only silly, but also makes you look like a huge tool.
7.
by rsanheim 5mo ago
Wow. So much hate in the comments here. Of all the funding / equity events lately, I wonder how this one gets so much doubt and distrust from the start. If this isn’t something to at least root for, in the sense of a small team, novel
8.
by rsanheim 6mo ago
Yeah /simplify is your friend. That and constrained prompts - “refactor x for simplicity - resulting diff must remove n lines of code. Dont change tests. “
9.
by rsanheim 6mo ago
Wait what. For real? I knew their security posture was bad, but this bad??
10.
by rsanheim 6mo ago
This has been the case for 15-20 years at least. It’s only now that the horrible experience for regular users is so obvious compared to Linux becoming quite good, and Mac OS ranging from fine to meh. The continual recall/ai push from M
11.
by rsanheim 6mo ago
If you believe this has any long term staying power at MSFT, I have a bridge to sell you paid for with MSFT stock funded by quarterly earnings reports.
12.
by rsanheim 6mo ago
Yes. This. Too little too late MSFT. And it really comes down to $MSFT. If the stock keeps dropping, how long do you think any real commitment to “quality” for a boring, low(no?) revenue product will last? Very little when the ad/partn
13.
by rsanheim 6mo ago
I’ve found LLM development expands the scope of what I can do to an absurd level. This is what exhausts me. My limits are now many of the same things that are have always been core to software dev, but are now even more obvious: - what is t
14.
by rsanheim 6mo ago
Codex does something sorta magical where it auto compacts, partially maybe, when it has the chance. I don’t know how it works, and there is little UI indication for it.
15.
by rsanheim 7mo ago
No. Nope. Agents do just fine with all sorts of CLIs. Old standards, new custom stuff, whatever. The CLIs I’ve seen agents struggle with are those that wrap an enormous, unwieldy, poorly designed API under one namespace. All of Google Works
16.
by rsanheim 7mo ago
he favors Codex? The original name of his ai assistant tool was 'clawdbot' until Anthropic C&D'ed him. All the examples and blog posts walking thru new user setup on a mac mini or VPS were assuming a claude code max accou
17.
by rsanheim 8mo ago
> But on the nuclear issue, it's not a good sign that he's explicitly saying that this AGI future is a threat to nuclear deterrence and the triad. Like, where do you go up from there? That's the highest level of alarm that
18.
by rsanheim 8mo ago
Also many CLIs act differently when invoked connected to a terminal (TUI/interactive) vs not. So you’d run into issues there where Claude could only test the non-interactive things.
19.
by rsanheim 8mo ago
Is there a term for AI-fueled dev psychosis? "AI architecture astronaut" ? There should be one if not. Or maybe just AI-fueled hucksterisim... I recognize 100% that a tool to manage ai agents with long term context tracking is goi
20.
by rsanheim 8mo ago
The performance of hardware today is even more mind-boggling compared to what most people (SRE managers, devs, CTOs) are willing to pay for when it comes to cloud compute. even more so when considered in the context of dev 'remote work
21.
by rsanheim 9mo ago
These exist, typically made by Panasonic or Sony, and cost upwards of 20k USD. HDTVtest has compared them to the top OLED consumer tvs in the past. Film studios use the reference models for their editing and mastering work. Sony specificall
22.
by rsanheim 9mo ago
_low_type_ is early days still, but I think this approach is clearly the future of ruby typing. If this gets baked into the language for full “compile” time support and minimal performance impact, it will be amazing: https://gith
23.
by rsanheim 9mo ago
Yeah, this is very real, and I think it can inflict paralysis on programmers with a certain level of experience and 'i know better' syndrome. Or even a 'it _might_ be better' type syndrome. Sometimes, you might really kn
24.
by rsanheim 9mo ago
It was really a blending of multiple genres. Was really ground breaking by for its time.
25.
by rsanheim 10mo ago
But …you have to give the MCP the creds somehow. Maybe it’s via a file on disk (bad), maybe via an env var (less bad). Maybe you do it via your password CLI that you biometricly auth to, which involves a timeout of some sort for security, b
26.
by rsanheim 10mo ago
I think the answer is it doesn't fit in any definition of a _good_ monitoring stack, but we are stuck with it. It has largely become the blessed protocol, specification, and standard for OSS monitoring, along every axis (logging, traci
27.
by rsanheim 10mo ago
Did not feel LLM written to me - at least not overtly so. LLM editing/assisted perhaps? It was a fun little post that felt accurate (ie confirmed my own biases ;)) about the current state of LLM models in a silly, but real, use-case. T
28.
by rsanheim 11mo ago
because its a huge hassle that many financial services companies have no incentive to facilitate or make easy or discoverable. And for many folks a job change is a stressful event even in the _best_ of circumstances. I know when I was laid
29.
by rsanheim 11mo ago
Just...don't use them? I've use mise happily for many months without using direnv or tasks, and everything I use it for works and is solid. Installs python, ruby, node, does the switching, does the shims, stays out of the way. dir
30.
by rsanheim 11mo ago
I wonder what kind of outage or incident or economic change will be required to cause a rejection of the big commercial clouds as the default deployment model. The costs, performance overhead, and complexity of a modern AWS deployment are i

More ›