maherbeg

I don't want unlimited writes. I basically want to unlock nearly everything but approve writes in some scenarios.

lol yeah, but mostly just want to allow more types of reads for getting context, and primarily for test running / linting etc. I shouldn't have to approve every invocation of `pytest` or `bazel test`.

Yeah this has been my experience as well. The Claude Code UI is still so much better, and the permissioning policy system is much better. Though I'm working on closing that gap by writing a custom policy https://github.com/openai/codex/blob/main/codex-rs/execpolic...

Kinda sick of Codex asking for approval to run tests for each test instance

It's both context and memory. If an LLM could keep the entire git history in memory, and each of those git commits had enough context, it could take a new feature and understand the context in which it should live by looking up the history of the feature area in it's memory.

Yeah, this is absolutely true. It can be configured to do anything, which means you really need to make sure you've configured everything correctly.

huh, I don't seem to have this enabled at all. My cd in fish only shows the current directory choices

These would be great for outdoor running to keep tabs on your pace, HR, time etc. Paired with the new airpods 3, you could be phone free and watch free.

The one downside to nix that I've found is that nix rebuilds with flakes are incredibly slow. I'm using determinate nix too on nixos and it's still sooo slow

Yes, but that final claude is much further away than people think. So for a while, enhancing human productivity seems like a benefit

etcd is raft, but zookeeper is paxos as a service. so maybe a hosted version since zookeeper can be troublesome for folks to operate.

There's a sweet omarchy-nix config that I really like https://github.com/henrysipp/omarchy-nix

I highly recommend having fairly succinct project level CLAUDE.md files, and defer more things into sub-folders. Use the top level as a map. Then during your planning of a feature, it can reach into each folder as it sees fit to find useful context to build out your phased implementation plan. I have it use thinking mode to figure out the right set of context.

At the end of each phase, I ask claude to update my implementation plan with new context for a new instance of claude to pick it up. This way it propagates context forward, and then I can clear the context window to start fresh on the next phase.

People are more insulin sensitive in the morning too which could correlate with this.

a nice way to solve this is a hash ring and put it in the ID. you can re-shard the ring as you need to, and can route to where the object lives easily.

instance store on aws can give up to 3.3mil iops https://aws.amazon.com/blogs/aws/now-available-i3-instances-... - the main problem is just using networked storage.

Seems pretty guilty to me if you're going to make an escape somewhere you can't get extradited.

The biogents traps are really effective. We have one without CO2 on our property and it works amazingly.

The blog post mentioned correlated failures in an availability zone. You likely could reduce this a bit, but still run into this frequently enough

Why don't insurance companies mandate significant fire abatement in new builds to be insurable? While it may not be possible to save every house, I wonder how many houses could have been saved with a thought behind "how can we minimize damage in a wildfire scenario"

You should also try using Libby / Overdrive to see if your local library has the book. I've borrowed 20+ books this year and it's pretty seamless and easy!

It says that a color image is 150ppi, and black and white is 300ppi. How does a colored highlight work on a mostly b&w page? does that section have reduced sharpness?

There’s been another hypothesis around the wider spread use of emulsifiers, specifically in lower fat foods. Lots of people are eating “healthy” almost whole food sources that contain them.

I'd be curious to know how it is different than Amethyst. Amethyst has gotten much more stable recently and I really enjoy using it. The text based config seems pretty nice at a first glance, but I'm not sure if it's worth switching.

True, but each advance adds up. Just look at a history of curing hodgkin's lymphoma which omits a lot of detail still https://www.hematology.org/about/history/50-years/milestones...

I think most 2 car households could have a full EV and a hybrid combination and basically never use gas throughout the year.

We moved to 2 full EVs and have no problems getting anywhere (Texas to Wyoming to Illinois and back was easy)

They always were given that is a part of their mission.

This could also be a good thing in that they may not be even close to real AGI.

I'm curious what the fix was, presumably just retry?

I switched to using Zed recently and quite like it. There are definitely teething problems dealing with all of the different user and tooling configurations out in the wild, but it does feel significantly faster than VSCode. I'm only missing a few things from VSCode (like inline test runners and bazel plugins). I'm looking forward to their take on dev containers / remote dev too.

I hope they survive as a company!

You can also install solar tubes to do something similar.