HNHacker News
TopNewBestAskShowJobs

mroche

3,592 karma · joined October 4, 2018

Systems Engineer at Skydance Animation

https://omenos.dev/

Online Accounts:

Git{Hub,Lab}, Codeberg, SourceHut: @omenos

Bluesky: @omenos.dev

Matrix: @mroche:fedora.im

Mastodon: @omenos@fosstodon.org

submissionscomments
mroche··on Sonnet 5.5
Is there ever any focus on producing new Haiku models? There are a lot of use cases for quick to return models when you're limited to a single provider.
mroche··on DEA to Temporarily Schedule 7-Oh and Related Substances to Protect Public Safety
I know very little of this but it seems like not all things kratom are affected.

> This temporary scheduling action does not apply to botanical kratom products that contain naturally occurring 7-OH below the specified threshold. Instead, it targets synthesized products and those containing elevated concentrations of 7-OH as outlined in the temporary scheduling order. DEA believes these substances pose an imminent threat to public safety given their effects are highly unpredictable.

mroche··on Kagi Changelog (July 2): Heads, tails, and an AI toggle
I'd been hoping for G4 to be added so I could play around with it. Pretty much spent all of last week doing so. I thought something was short-circuiting due to the speed of response, I'd never seen a model operate that fast.

https://inference-docs.cerebras.ai/models/gemma-4-31b

mroche··on Computer use in Gemini 3.5 Flash
The graph has Gemini 3.5 Flash matching Sonnet 4.6, losing to Opus 4.8, and slightly behind GPT-5.5 by 0.3 points... That's not that much of a hands-down loss for Gemini for this specific workload benchmark.

The methodology used:

https://deepmind.google/models/evals-methodology/gemini-3-5-...

Methodology: All Gemini scores are pass @1 except where otherwise noted. "Single attempt" settings allow no majority voting or parallel test-time compute. All of the results are all run with the Gemini API for the model-id gemini-3.5-flash with default sampling settings unless indicated otherwise below. To reduce variance, we average over multiple trials for smaller benchmarks.

All the results for non-Gemini models are sourced from providers' self reported numbers unless otherwise mentioned below. For Claude Opus 4.7 , Sonnet 4.6, and GPT-5.5 we default to reporting maximum thinking/reasoning settings available, but when reported results are not available we use best available reasoning results.

mroche··on .gitignore Isn't the only way to ignore files in Git
Browsers are some of the most annoyingly configured applications out there.
mroche··on .gitignore Isn't the only way to ignore files in Git
You may need to have certain directories be excluded depending on the programs you use. For example, the default Chrome profile location is within ~/.config, which includes cache data that can be multiple gigabytes in size.
mroche··on Uv is fantastic, but its package management UX is a mess
This sounds... not possible for the core problem of how Python handles dependency resolution during the life of an application? How are you setting things up so that the following scenario is valid?

    program
    ├── dependency_a
    │   └── dependency_c (1.0.0)
    └── dependency_b
        └── dependency_c (2.0.0)
Otherwise, you've created a magic layer hack to enable multi-version dependency chains in a mono-version dependency chain language.
mroche··on Zig → Rust porting guide
I do not know if there's any overlap between these teams, but it seems like Anthropic itself is fairly invested in the Rust ecosystem.

They recently proposed some of their internal tools to be the official Rust implementation[0] of Connect RPC[1]. As a protobuf based library set, this includes a new Rust-based protobuf compiler, Buffa[2].

[0]: https://github.com/orgs/connectrpc/discussions/7#discussionc...

[1]: https://connectrpc.com/

[2]: https://github.com/anthropics/buffa

mroche··on Backblaze has stopped backing up OneDrive and Dropbox folders and maybe others
My understanding of Backblaze Computer Backup is it is not a general purpose, network accessible filesystem.[0] If you want to use another tool to backup specific files, you'd use their B2 object storage platform.[1] It has an S3 compatible API you can interact with, Computer Backup does not.

But generally speaking, I'd agree with your sentiment.

[0]: https://www.backblaze.com/computer-backup/docs/supported-bac...

[1]: https://www.backblaze.com/docs/cloud-storage-about-backblaze...

mroche··on Go hard on agents, not on your filesystem
The more you know, thanks for the information!
mroche··on Go hard on agents, not on your filesystem
> The claude sandbox is a good idea, but to be effective it would need to be implemented at a very low level and enforced on all programs that claude launches.

I feel like an integration with bubblewrap, the sandboxing tech behind Flatpak, could be useful here. Have all executed commands wrapped with a BW context to prevent and constrain access.

https://github.com/containers/bubblewrap

mroche··on AMD's Ryzen 9 9950X3D2 Dual Edition crams 208MB of cache into a single chip
I'm looking at building a new system, and was waiting to see what happens with this chip and Intel's Arc Pro B70 card. I can't find ECC UDIMMs of 64GB per-stick to make 128GB, but I can put together two solo UDIMMs of 32GB or 48GB for $800 and $1000 per stick respectively.

I really want to see what enabling the L3 cache options in the BIOS do from a NUMA standpoint. I have some projects I want to work on where being able to even just simulate NUMA subdivisions would be highly useful.

mroche··on A Faster Alternative to Jq
> Having used `jq` and `yq`

If you don't mind me asking, which yq? There's a Go variant and a Python pass-through variant, the latter also including xq and tomlq.

mroche··on Mystery jump in oil trading ahead of Trump post draws scrutiny
> TACO

TIL: https://en.wikipedia.org/wiki/Trump_Always_Chickens_Out

mroche··on Tin Can, a 'landline' for kids
Y-Zer myself and I do the same thing. I never initiate the communication when called unless I am expecting it or I know who the caller is. Otherwise, they'll know when someone picked up because their side will stop ringing, and they'll only get awkward silence until they start talking. Often times it's an automated voice system that will not begin until prompted by the callee, so it hits a timeout and hangs up.

The number of calls I get where it's either dead silence in the other end or clearly a call center based on the noise can only be categorized as "too much".

mroche··on MAUI Is Coming to Linux
What's needed beyond API docs is a review, refresh, and possible merging of the two Wayland Books by active Wayland contributors.

https://wayland.freedesktop.org/docs/book/ https://wayland-book.com/

mroche··on Standardizing Pure PQC
First time I've heard of SCW...

Of all the AI voice-generated things I've heard so far, this definitely takes the cake for the funniest tech-related conversation.

mroche··on FP-Go V2: Enhanced Functional Programming for Go 1.24
fp-go/v2 was recently released in Dec 2025.

Prior discussion around fp-go/v1 in 2023:

https://news.ycombinator.com/item?id=37171149

mroche··on Ask HN: Please restrict new accounts from posting
Your comments use em dashes. Many would claim those are vastly overrepresented in AI language and thus an account overly using them are blatantly AI.

I've always found this funny. Doesn't macOS' default text substitution enable (annoying to me) things like em-dash, smart quotes, etc?

mroche··on Bootc and OSTree: Modernizing Linux System Deployment
It is not, the future is currently pointing to composefs:

https://github.com/bootc-dev/bootc/issues/1190

There's a GitHub org that builds bootc-ready images for non-Red Hat family distributions using this backend.

https://github.com/bootcrew

mroche··on New accounts on HN more likely to use em-dashes
> super-comma

This is the first time I've ever heard the character ";" referred to as such. It's always been "semi-colon" to me, is this a region/culture difference?

I'm not saying you're wrong, I find it interesting.

mroche··on I found a useful Git one liner buried in leaked CIA developer docs
Likewise with the other way around, just switch pull with push.
mroche··on Paged Out Issue #8 [pdf]
From PO's submission FAQs and policies:

> Obviously the used fonts should be readable (and ideally their name shouldn't start with "Comic" and end with "Sans", though there might be some article topics that justify even that!), and while almost any font meets this requirement, please be careful when selecting a non-standard font.

I kinda want to see such an article, but taken seriously discussing the history of the font, its design and purpose, evolution, and purpose-related/derivative font families.

mroche··on Losing 1½ Million Lines of Go
> Unfortunately, Go’s library doesn’t get updated every time Unicode does. As of now, January 2026, it’s still stuck at Unicode 15.0.0, which dates to September 2023; the latest version is 17.0.0, last September. Which means there are plenty of Unicode characters Go doesn’t know about, and I didn’t want Quamina to settle for that.

I have to say I am surprised about that. Does anyone have any context or guesses as to why this is the case?

EDIT: Go's unicode was actually updated to v17 yesterday:

https://github.com/golang/go/commit/dd39dfb534d2badf1bb2d72d...

mroche··on Personal thoughts/notes from working on Zootopia 2
A similar approach was taken by Pixar when making the beach environment for the Piper short that was previewed with Finding Dory. It was absolutely mind-blowing.

https://www.fxguide.com/fxfeatured/the-tech-of-pixar-part-1-...

Related but unlinked Part 2 on other aspects of Finding Dory:

https://www.fxguide.com/fxfeatured/the-tech-of-pixar-part-2-...

mroche··on OpenGitOps
Would you mind elaborating on this more, describing the differences and how tools like Zuul introduce degrees of friction that result in smooth operation and pipelines?

I know my phrasing may come off wrong, I apologize for that. But I'm asking genuinely; I've only ever seen Zuul in the wild in the Red Hat and OpenStack ecosystems.

mroche··on Archiving Git branches as tags
That's the purpose of the forge platform, to provide a way to prevent changes to these files from being accept into the source repository. For example:

https://docs.github.com/en/repositories/configuring-branches...

https://docs.gitlab.com/user/project/protected_tags/

https://forgejo.org/docs/latest/user/protection/#protected-t...

mroche··on Archiving Git branches as tags
I 100% agree on the latter (the tag != release is more of a project management issue), and the same concept applies to containers and their digest hashes. The main issue at the end of the day is the human one: most people don't like looking at hashes, nor do they provide context of progression. I would say "give both" and make sure they match on the end user side of things, but tags are the most common way (open source) software releases are denoted.
mroche··on Archiving Git branches as tags
The purpose of the forge is to be able to prevent this. Protected tags are usually a feature which provides a way to mark tags as untouchable, so removal would require a minimum level of trust to the repository on the platform. Otherwise, attempts to push tag deletions or changes for tags matching the protected pattern would be rejected/ignored.

Of course, the repository owner has unlimited privilege here, hence the last part of my prior comment.

mroche··on Archiving Git branches as tags
Luckily commonly used forges for collaboration have the ability to make tags immutable. Any repository where multiple people collaborate on a project should have that feature enabled by default. I'm still waiting for the day where tags are immutable by default with no option exposed to change it.

I'm sure that would cause problems for some, but transitive labels already exist in Git: branches.

Page 1 of 17Next →