mberning

Interesting. I am working on a project to integrate access management to iceberg/parquet files for sagemaker. Controlling what users logged into sagemaker studio have access to in s3. It’s fine using static policies for mvp, but eventually it needs to be dynamic and integrated into enterprise iam tools. Those tools generally have great support for managing sql grants. Not so much for s3 bucket policies.

Does anyone know how access control works to the underlying s3 objects? I didn’t see anything regarding grants in the docs.

The Napier Deltic has a very distinctive sound. You can hear it in locomotive form on youtube. If you are into that sort of thing there are some really good videos on the Rolls Royce Crecy engine as well.

Awesome story. Thanks for sharing.

Sounds like your homelab became too similar to a Fortune 500 IT job.

I think they have also done studies in rodents that show pm2.5 diesel particulate decreases insulin sensitivity.

If advertising didn’t work to drive consumption no business would ever spend a dime on it. When your product has addictive qualities advertising absolutely increases the number of people that get hooked. This is something that was well understood in the past, for example the huge curtailment in cigarette advertising.

I don’t have an overly paternalistic view of the government. I’m rather libertarian in that regard. But is it too much to ask that we place some guardrails on things that are know to have trouble with? Smoking, drinking, gambling, etc.

I certainly feel that people should be able to do it if they really want to, but making it super accessible and highly advertised seems like a bad idea.

It’s expensive and the “design” of the services, if you could call it that, is such that you are forced to pay a lot, or play a lot of games to get around it. If you are going to spend your engineering time working around their ridiculous pricing schemes, you might as well spend the money on building things out yourself.

Perfect example - MSK. The brokers are config locked at certain partition counts, even if your CPU is 5%. But their MSK replicator is capped on topic count. So now I have to work around topic counts at the cluster level, and partition counts at the broker level. Neither of which are inherent limits in the underlying technologies (kafka and mirrormaker)

Having a distinctive style is also going to become increasingly rare in the age of LLMs. I see it with people at work already. They have no style. They learn very little. They get something to work but have no ability to discern whether it is a “good way” of doing it or simply “a way” of doing it. They have little interest in how or why things are built. I get the sense that they have little or no enjoyment or interest in deep understanding.

As a concrete example I just reviewed a PR for a feature that someone wanted to add to a flask app. It had a ton of terraform code, an aws api gateway, a lambda, etc. then on the flask app frontend they added a page with a call to this new API. I asked why they didn’t just add a route to the flask app. Blank stare response.

In the US it is very easy to test your vitamin D levels. I recently had mine done and was just below normal range. Started supplementing and will test again in 6 months.

They have certainly tried to move the goalposts on this.

I do something similar. I create a bare repo on my dropbox folder or nas mount. Then checkout from bare repo file path to some place where I will be doing all the work.

Back up for me now.

CICS and HATS are perhaps the most annoying pieces of technology I’ve ever encountered.

If we could get circa 2010 Google back people would not waste nearly as much time with “AI”.

At around the age of two my son became totally obsessed with construction equipment. We would pull over and let him watch excavators and skid steers work at local job sites. Luckily there are some good youtube channels that cater to this which made it much easier to scratch his itch. He still loves heavy machinery, but at the age of 5 it’s a lot more mellow.

I agree 100% on the added cost and complexity. I think a lot of that gets masked by the boilerplate that makes it so easy to get started. Then they have their hooks in you and 6 months or a year later you are scratching your head and hiring “react developers” to help solve the problems you were trying to avoid.

I don't know what point you are trying to make. One person's "abusive and toxic work environment" is another's "results driven organization". If you find yourself in a role that doesn't fit you, or you have serious doubts about, my best advice would be to leave. You are unlikely to meaningfully change the culture. In fact it will probably achieve the opposite of what you want, and people will dislike you for trying.

That presumes you know all secrets ahead of time. A risk in and of itself. But from a practical point of view you will never know all secrets, because they are generated constantly in real time.

Tooling goes way beyond the editor/IDE. Eclipse is a very good free option. As is IntelliJ CE. I personally have the all products pack and use the ultimate version.

Beyond the IDE you also have to consider the build tools, package management, debuggers, profilers, static analysis tools, etc.

It’s honestly too much for an HN comment. But as an example, if I do open one of these awful projects at work and it uses gradle for example, intellij will understand that, import the project, get all dependencies, let me run any target with debugging or profiling, give me code coverage, etc.

I’ve worked in Java for over 20 years. Being the “lingua franca” of the enterprise is its biggest strength IMO, but it is also perplexing to me that it was able to do that in the first place. The language itself is not bad. The tooling around it is very good. But the codebases you encounter written in it, particularly in the enterprise, are often horrible.

This is basically code for only wanting to hire young people with little or no commitments outside of work. As a bonus they usually undervalue themselves.

If you work at a real enterprise that actually takes security seriously I can assure you a large portion of it is not theater. You will find this out when they come knocking and point out something boneheaded that happened on your watch. I once had an intern that mistakenly committed a non-prod credential into source control. They realized their mistake and replaced it with a token. But not before it had triggered some infosec alert and they blasted me with a stern “ACTION REQUIRED” email. I also had people on my team get snagged by simulated phishing emails and other such things which are run constantly.

The only thing it’s upending is sane engineering practices in my experience. I just reviewed a python flask app which was basically “vibe coded” by a couple people speed running a checklist of features. I open up the auth.py file to find it is integrated with an idp using oauth. Whether the caller has the correct scopes is never checked. I grab client credentials for a completely unrelated app. Hit /admin using my bearer token. Sure enough anybody with a valid set of client credentials could have had full admin access to this app for a year. But hey, at least when I asked how they tested the app they were quickly able to generate thousands of lines of pytest.

What is the use case for this? Their cloud gateways are very affordable. Aside from the academic aspects of it I don’t see why I would want to run a home brew version of their cloud gateway.

I am highly sympathetic to all impacted. It’s devastating.

The rhetoric around cost cutting is something I can’t figure out how to deal with. The media makes it sound like the sky is falling with every cut. But having worked in enormous organizations myself, I also know that teams can and do lose headcount all the time, and somehow manage to carry on.

I guess the issue I wrestle with the most is how we can spend tax dollars on programs, but also have the ability to reduce those programs when needed, without it seeming like the end of the world.

We can’t have government agencies and programs that go on forever, expanding, growing in headcount and budget, for eternity. But the rhetoric around these events would have you believe that is the only path forward.

I think photo booth had the most depressing “evolution”.

Agent mode in RubyMine which I think is using a recent version of sonnet. I tried starting a new agent conversation but it was still off quite a bit. For me my interest in finessing the LLM runs out pretty quickly, especially if I see it moving further and further from the mark. I guess I can see why some people prefer to interact with the LLM more than the code, but I’m the opposite. My goal is to build something. If I can do in 2 hours of prompting or 2 hours of doing it manually I’d rather just do it manually. It’s a bit like using a mirror to button your shirt. I’d prefer to just look down.

Same. I had some idea that I wanted to build a basic sinatra webapp with a couple features. First version was pretty good. Then I asked it to use tailwind for the css. Again pretty good. Then I said I wanted to use htmx to load content dynamically. Suddenly it decides every backend method needs to check if the call is from htmx and alter what it does based on that. No amount of prompting could get it to fix it.