HN user

robbintt

406 karma
Posts1
Comments231
View on HN

That is pretty compelling. Probably another layer you use by default but can swipe to the gui search layer or full control. As usual, nerds will default to full control.

I have 1 of these for work and 1 for home. I'm waiting 5 years from release, so about 2024-2025. Computers got really good around 2017, and the only reason to upgrade is the heat and fan. I use remote VMs anyways, so chrome is really my limiter.

It definitely reduced the amount and complexity of work there is to do. So, in a static market, it reduced the number of workers. It's hard to see in tech which is an expanding market.

I went to a tech presentation by Barrick and got the impression that gold price is set by their fixed costs and operation costs. If the price goes up they turn up their operations. I am sure this isn't a controlling factor as demand could outstrip their operation size. But it does set a soft ceiling when demand is low, and it is.

The general thinking about this is to use a "safe withdrawal rate" of 3% to 4% of assets per year. Dividends are out of fashion since they don't offer tax deferral, they also aren't really easy to properly diversify although dividend funds are offered.

I have asked up front if people attach timelines to their offers or use exploding offers and just decline to interview. Those offers are, in my experience, always less competitive.

It's not negligible, BECAUSE you are seeing dozens of patients a day. When you're a doctor, factoring the odds into your diagnosis will screw up your diagnosis BECAUSE you are exposed to the whole probability spectrum, not a quanta of it.

Using "chance of occurrence" is a major logical fallacy for diagnosis identification and can ruin the whole practice.

I completely agree that it has problems. I use terraform a lot. Compared to nothing, I love terraform. However, it's so overwrought that it has to be hidden from anyone not in infra for a living.

1. IaC description should be format agnostic and transformable (eg definable in yaml, json, whatever).

2. Something about provider interfaces here, but it's already super messy and not sure if it's an improvement or just a shift

3. State files were wild west last time I checked. And there should be a default database interface provider at minimum. Maybe there is now?

4. Forcing the apply->statefile cycle as the default requires all of compute, interface, and a human. This should have been an abstraction on a raw interface for automated use.

It will affect categories of business users (programmers) who currently embrace terraform: amazon, google, microsoft, oracle, alibaba cloud.

Like it or not, cohorts of engineering organizations like the above (cloud providers) have a very outsized weight and already have contender products they can choose to vigorously fund tomorrow.

From the article:

The license does not allow you to use Terraform if you meet both of the following conditions: You are building a product that is competitive with HashiCorp. You embed or host Terraform in your product.

My $0.02: the management of hashicorp is following a stupid trend and should have thought about their customers more.

It will come out to what lawyers think, I guess. Lawyers usually say no to things with poorly established precedent.

To use the apt analogy further up in the thread, `sudo apt install git` is repeatable in your dockerfile, but often not reproducible. Later on you will get a different build. Across say 500 packages and 1,000,000 containers (or say 1000 container images if you are deploying images) over even a week this becomes extremely... varied...