carderne

What do you gain by having multiple repos?

John McPhee talks about a similar model for the Mississippi River in “The Control of Nature” Well worth a read. Fun stories about Hawaii and Los Angeles too, iirc.

The UK grid does not import 44% of its energy.

Gell-Mann suggests I should treat the rest of your post with skepticism.

EDIT: maybe you meant the UK total primary energy? I feel like that's extending the boundary a bit far. Should we start digging for uranium? Or stick to renewables, but only with locally sourced silica and rare earths?

What API/approach does it use to edit sheets?

I made a CLI (+skill) so agents could edit files with verbs like `insert A1:A3 '[1,2,3]'`, but did some evals and found it underperformed Anthropic's approach (just write Python).

I deliberately didn't mention this because I think most of the pain with Github over the last year is probably caused to some degree by their scale, which seems like an unrelated issue. (But maybe not.)

If anyone from Astral sees this: at this level of effort, how do you deal with the enormous dependence on Github itself? You maintain social connections with upstream, and with PyPA... what if Github is compromised/buggy and changes the effect of some setting you depend on?

I’m surprised it works for you with such a simple config? I’m the one that added the allowRead option to Claude’s underlying sandbox [0] and had quite a job getting my toolchains and skills to work with it [1].

[0] Fun to see the confusing docs I wrote show up more or less verbatim on Claude’s docs.

[1] My config is here, may be useful to someone: https://github.com/carderne/pi-sandbox/blob/main/sandbox.jso...

Interesting, that's not been my experience! Maybe you've got the list of things to allow/block just right. While testing different policies I've frequently seen Opus 4.6 go absolutely nuts trying to get past a block, unless I made it more clear what was happening.

Yeah I think for general use the transparency of what your thing does is really great compared to a pile of TypeScript and whatnot.

How do agents tend to deal with getting blocked? Messing around with sandboxes, I've quite even seen them get blocked, assume something is wrong, and go _crazy_ trying to get around the block, never stopping to ask for user input. It might be good to add to the error message: "This is deliberate, don't try to get around it."

For those using pi, I've built something similar[1] that works on macOS+Linux, using sandbox-exec/bubblewrap. Only benefit over OP is that there's some UX for temporarilily/permanently bypassing blocks.

[1] https://github.com/carderne/pi-sandbox

There is sandbox-runtime [1] from Anthropic that uses bubblewrap to sandbox on Linux (and works the same as OP on macOS). You can look at the code to see how it uses it. Anthropic's tool only support read blacklist, not a whitelist, so I forked it yesterday to support that [2].

[1] https://github.com/anthropic-experimental/sandbox-runtime [2] https://github.com/carderne/sandbox-runtime

I guess you don't get the luxury of being opinionated enough to say: forget your old data.

Anyway great write-up, even though I'm sure it's painful having to run this system on top of your once-elegant Postgres solution.

How does this work with self-hosting? Is the assumption that self-hosters won’t run into this problem?

For most use-cases I’d probably prefer to just delete the payloads some time after the job completes (persisting that data is business logic problem). And keep the benefits of “just use Postgres”, which you guys seem to have outgrown.

I got pi to write me a very basic sandbox based on an example from the pi github. Added hooks for read/write/edit/bash, some prompts to temp/perm override. Have a look, copy-paste what you like.

https://github.com/carderne/pi-sandbox

The people pushing oh-my-pi seem to have missed the point of pi... Downloading 200k+ lines of additional code seems completely against the philosophy of building up your harness, letting your agent self-improve, relying on code that you control.

If you want bags of features, rather clone oh-my-pi somewhere, and get your agent to bring in bits of it a time, checking, reviewing, customising as you go.

I'd say it's the idea/fact/feeling that, in 2026, agency matters more than skill/wisdom/intelligence.

Long read on the topic (quite funny, covers Cluely): https://harpers.org/archive/2026/03/childs-play-sam-kriss-ai...

It means the marginal cost to sell another subscription is lower than what they sell it for. I don't know if that's true, but it seems plausible.

This is super interesting framing. I’m definitely a completer, not that I like much about Slack. Probably useful to have this kind of discussion before/while making knowledge management decisions in startups.

Out of interest, what control plane do you use for a Hetzner/metal setup? Kubernetes ecosystem?

I use Coolify for side projects, haven’t investigated whether I’d want to use it for bigger/importanter stuff.

I wasn't able to find any imagery online, and I don't have anything I can share publicly.

These are some of the existing commercial solutions (just found these on Google, can't remember which I was comparing my own work against):

- https://koregeosystems.com/digital-core-logging/

- https://mountsopris.com/wellcad/core-logging-software/

- https://www.geologicai.com/logging/

I don't know enough about the science side to take it any further on my own.

The "tech" part of what I started building is really quite simple: convert the images to Cloud-optimised GeoTIFF, then do range requests to S3 from the browser.

Might not be possible to find any, they’re expensive and niche. If you reach out (email in profile) I can show/share how it works (nothing currently public).

I did something similar once for a mining technique called “core logging”. It’s a single photo about 1000 pixels wide and several million “deep”: what the earth looks like for a few km down.

Existing solutions are all complicated and clunky, I put something together with S3 and bastardised CoGeoTIFF, instant view of any part of the image.

Wish I knew how to commercialise it…

To be fair, your comment didn't add much either.

Their main criticisms of Python were:

> it is slow, its type system is significantly harder to use than other languages, and it's hard to distribute

Your comment would have been more useful if it had discussed how FastAPI addresses these issues.

I’m just a casual observer of this thread, but I think you’d find it worthwhile to read up a bit on zero-copy stuff.

It’s ~impossible in Python (because you don’t control memory) and hard in C/similar (because of use-after-free).

Rust’s borrow checker makes it easier, but it’s still tricky (for non-trivial applications). You have to do all your transformations and data movements while only referencing the original data.

peewee is nice, but I can't write un-typed Python anymore, so it's not an option for me.

I was hoping the parentheses themselves would be flipped. Like this:

> 1 + )2 * 3(

(1 + 2) * 3

Users of a library will generally instruct their type-checker not to check the library.

So only the outer API surface of the library matters. That’s mostly explicitly typed functions and classes so the room for different interpretations is lower (but not gone).

This is obviously out the window for libraries like Pydantic, Django etc with type semantics that aren’t really covered by the spec.

Embar: https://github.com/carderne/embar

A Python ORM, inspired by Drizzle and the like. Whenever I come to Python I'm frustrated by the ORM options. They generally lack type-safety on inputs and outputs, or useful type hints.

SQLAlchemy is an institution but I think it's hard to use if it's not your full-time job. I check the docs for every query. I want something simple for the 80-99% of cases, that lets you drop easily into raw SQL for the remaining %.

I'm going to keep hacking at it, would love to from anyone who thinks this is worthwhile (or not). Also: - The interface for update queries is clunky. Should I add codegen? - Should I try to implement a SQL diffing engine (for migrations). Or just vendor sqldef/similar...?

Is it absolutely necessary to have a base/gateway? This Verge article[0] seems to imply not, but it's not at all clear to me what I lose.

If I just want a smart switch that controls a smart light, can I do that without a hub? Can I use my phone to control that light/switch in a pinch? I'm not averse to spending $100 or whatever, but it's just more _stuff_ that I'd rather not think about.

[0] "Apple now lets you add Matter devices to Apple Home without a hub" https://www.theverge.com/2024/9/18/24246581/ios18-matter-sma...

What “hub” thing do I need to use this? I use an iPhone but really just want to use some physical remote switches.

I currently use Home Assistant but want to shift to something more “mass market” as I’m bored of being family tech support.

Here it is for anyone else who struggled:

https://support.mozilla.org/en-US/kb/profile-management