An articulated rationale for why a fork would go a long way for awareness of why an existing user of Helix might consider to use Mitos instead (or not) really needs to be the first thing we see on the front page of the site for Mitos.

One thing that really sticks out to me and gives me a feeling I won't consider to use Mitos, is the presentation further down on Mitos' front page of rendering of image files inside the terminal emulator. I rather that an image viewing program be opened, maybe called with options to not display interface widgets, and to be displayed in a new window tab. I use i3 or Sway, so I already have macros that do just this, and I don't fail to appreciate that others have this luxury (or not).

Alternative branches, or even forks like this are a good thing. Sometimes these help reach new consensus and merge back into the original project making the new project stronger, provided the original reason for a fork wasn't really issues of conflicting personalities.

I think a fork of Helix can be a step in the right direction. But "view images inside the terminal" doesn't pass the smell test. The mindset of displaying images inside of a terminal window and the macOS window decorations in the example screenshots, those two things together are part and parcel.

That's true, but I'm sorry, telling them to test it themselves every time a new release comes out because "I can't comment on known issues or on whether or when a fix might ship" is just stupid. It's a liability-reducing strategy benefiting only Apple, not a nice way to treat customers.

Also, not having a regression test that catches OpenGL being completely broken on many common hardware configurations seems pretty bad. Yes, OpenGL on macOS is deprecated, but guess what, deprecating it was also stupid. It was a legal decision made by Apple lawyers pursuing a particular litigation strategy vs Nvidia that caused Apple's withdrawal from Khronos group. It wasn't a sound technical decision.

Both of these things make it seem like the lawyers are in charge at Apple rather than the engineers.

A 1983 review of this remarkable system for the Institute of Electrical and Electronic Engineers (IEEE) attributes no fewer than eight "firsts" to SIGSALY. They are as follows:

1) The first realization of enciphered telephony

2) The first quantized speech transmission

3) The first transmission of speech by Pulse Code Modulation (PCM)

4) The first use of companded PCM

5) The first examples of multilevel Frequency Shift Keying (FSK)

6) The first useful realization of speech bandwidth compression

7) The first use of FSK - FDM (Frequency Shift Keying-Frequency Division Multiplex) as a viable transmission method over a fading medium

8) The first use of a multilevel "eye pattern" to adjust the sampling intervals (a new, and important, instrumentation technique)

From https://www.jproc.ca/crypto/sigsaly1.html

I think that the relevant patents remained classified until 1975.

Since this seems to get lost, I'd just like to point out: if you ask your agent to do something like this (or pretty much anytime you give it internet access and say "do whatever it takes"), you are responsible for what it does. Even if you couldn't have predicted the lengths to which it would go.

Intellectual property laws never totally made sense to me in the first place, but they make even less sense in the age of AI.

Who's "intellectual property" are we protecting, exactly, when a machine generates the knowledge?

I hope we can update before it's too late to matter.

First time I've seen "longer queues than the soviet union" used as a defence.

Every play made by these companies aligns with what you’d do if you have a $1T valuation to protect against an industry that seems increasingly commoditizable.

I’m still… hopeful… that “good enough” models will be abundantly available to be run anywhere as to limit the low water mark for market capture by a few big ones.

It’s also a “you can’t scrape us but we can scrape you” mentality these companies are getting into.

It really sucks for small businesses. Their websites get absolutely hammered and they can’t afford or don’t know how to stop it. Cloudflare ends up winning as they’re the easiest solution.

If you're on Python and your problem is too big for a single regex, but too small for a full-blown parser like in the article, I can recommend the built-in re.Scanner[0]. It's a regex-based tokenizer[1] (so no recursive parsing) in the `re` module, that for reasons is completely missing from any official documentation.

You give it a pattern for each token type, and a function to process each match, and you get back a list of processed tokens.

Importantly, it processes the list in one pass and ensures the matches are contiguous, where a naive `re.findall` with capture groups will ignore unmatched characters. You also get a reference to the running scanner, so you can record the location of the match for reporting errors.

    import re

    # Initialize the scanner.
    scanner = re.Scanner([
      (r"\d+",   lambda scanner, token: ('COUNT', int(token))),
      (r"\w+",   lambda scanner, token: ('ANIMAL', token.removesuffix('s'))),
      (r"[,. ]",     None), # None -> skip token.
    ])

    # Scan a sample string.
    results, remainder = scanner.scan("45 pigeons, 23 cows, 11 spiders.")

    # Ensure we parsed the whole thing.
    assert not remainder

    print(results)
    # [('COUNT', 45),
    #  ('ANIMAL', 'pigeon'),
    #  ('COUNT', 23),
    #  ('ANIMAL', 'cow'),
    #  ('COUNT', 11),
    #  ('ANIMAL', 'spider')]

[0]: https://stackoverflow.com/a/693818/252218

[1]: https://en.wikipedia.org/wiki/Lexical_analysis#Tokenization

None of the comments so far are engaging with the OP, whose written testimony was included in the article:

https://www.commerce.senate.gov/wp-content/uploads/media/doc...

It points to poorly designed and unproven "educational technology," or EdTech as the potential culprit. Rather than suggesting a luddite regression, it merely calls for proving the efficacy of EdTech prior to large rollouts.

Given the "reading wars" dragged on for over a century, I'm not optimistic this will be resolved anytime soon.

https://www.lexialearning.com/blog/the-science-of-reading-vs...

Regex is conceptually incredibly easy to understand.

But regex is also syntactically designed by Satan himself.

Most programming languages use commands that are semantically understandable, like "for" and "let" and "return" and mostly symbols we already know like "+" and "=". They use newlines and indentation for organization. And they put multiple commands on multiple lines and you can add comments and everything.

Regex is exclusively punctuation, with its own idiosyncratic meanings, all compressed into a single line, zero organization, zero comments, and impossible to remember what each symbol does if you don't use it daily. Not to mention the nightmare of escaping regexes when expressed as strings, on top of escaping regex symbols within the regex itself.

I swear it's high time we come up with a regex replacement that looks like normal Python code or HTML or something. So instead of:

  ^(\+\d{1,3}-?)?\d{3}-\d{3}-\d{4}$

We would write something like:

  <start>
    <optional>  # Optional country code
      "+"
      <digit>[1-3]
      "-"
    </optional>
    <digit>[3]  # Area code
    "-"
    <digit>[3]
    "-"
    <digit>[4]
  <end>

Same thing. But one is completely inscrutable to someone who hasn't learned regex, while with the other you can guess pretty easily how it works...

Figuring out how to make this was a pretty wild ride. I had zero clue how to approach this, so I went into the mountains with pencil and paper until I figured something out.

I started with a sort of "compile-time generational references" approach, where every reference would also remember when it was made, at compile-time. Then that evolved into a couple more things until I finally figured out this reverse approach. We'll see if it sticks.

Hope you all enjoy!

> this is a national bill introduced in the House

Nothing has been introduced–he has "announced he will introduce." It's still silly-season bluster.

To quote another reply regarding mermaid by @kepano:

> Mermaid.js has transitioned to "mermaid.ai" and is losing its original vision. Take a look at their new site.

> The latest version of Mermaid (12.1) ships with yet another layout engine (three now), and has a questionable safety record (see CVEs). In Obsidian 1.13 we had to introduce a new toggle/warning because of potential unsafe Mermaid diagrams.

https://x.com/kepano/status/2111575190081847776

> Enable JavaScript and cookies to continue

That about sums it up. Cloudflare has become the de-facto gatekeeper of the Internet, and that is IMHO very much NOT a good thing for anyone but Big Tech. Hostile user-agent discrimination to favour the incumbents, under the guise of "security".

> Suppose lean in its current state existed 15 years ago, and suppose the math community eagerly started contributing en masse to a shared repository of code. How much of the breakthroughs and longstanding questions in mathematics that we’ve seen answered this year do you think we would have solved?

(Former grad student in math here who has spent whole weeks formalizing in Lean lemmas that would be considered by professional mathematicians writing a mathematics text / paper as obvious ("without loss of generality we can assume that....", "it should be clear that...")

Probably very few or none.

These systems (mostly) don't help you discover new mathematics. They are a tool for formalization.

It just turns out that combining them with LLMs is the secret sauce. LLMs have quite a bit of creativity but make a lot of mistakes. The Lean typechecker never makes a mistake but has no creativity. Get them to work together in an iterative loop and you get what we have now.

Perhaps selfishly as I'm not much a gambler, I hate it because it ruined sports. You have a barrage of ads and odds that plague every game at every turn. You have a handful of players already caught betting on their own games, which means there are many more who just haven't been caught yet. And now, every single bad call leads fans to "refs must have money on this game", and to be honest, who knows if they do.

To clarify something that the OpenCode team has done a bad job of clarifying:

The leak was for something called 'OpenCode Console', which is distinct and separate from the OpenCode coding agent. If you're just using the coding agent, you're not affected. 'OpenCode Console' is a service that is similar to OpenRouter, but is much less widely used.

The name 'OpenCode Console' is super confusing, because one might think it means the CLI coding agent. It does not. 'OpenCode Console' has nothing to do with the CLI/TUI coding agent, nor the desktop app version.

The situation is made worse by the fact that 'OpenCode Console' barely has any documentation online, and the official leak announcement does not make the distinction clear at all. The OpenCode team did a terrible job of communicating this, and that's on them.

Today is a bad day for people using 'OpenCode Console'. But most people, probably the vast majority of OpenCode users who are reading this, are not using 'OpenCode Console'.

I'd been running into a lot of disk pressure on my system from agents running. I discovered two main causes:

- my approach to using git worktrees for all work. Not entirely surprising, the duplicated build artifacts in each worktree being the main culprit here. There were some limited gains to be had by just being more efficient with how builds were managed but ultimately I just had to limit the number of concurrent features I was working on in a couple of particularly large repos.

- more insidious was the fact I was actually hitting memory limits more than I realised and causing my mac to have to swap heavily. The system was surprisingly still quite performant while this was happening... until it wasn't. Unfortunately as best I can tell the only "fix" when this happens is to restart the machine. There's no way to have the system reclaim any of the swap space once it's written to disk. Which if you have it thrash a couple of times a day (especially easy if you leave something running overnight) adds up very quickly.

> Just glancing at the source you could do easy 100x performance improvements with simple fixes.

Now that you said this you really are expected to do a proof-of-concept implementation that is 100x faster, but with the same low-level concepts ;)

I think 2..3x might be achievable assuming the vibe-coded code quality is really badly optimized, but definitely not 100x.

As far as I understand the simulation, it does the entire address decoding and memory accesses as a transistor-level simulation (which is the whole point!), e.g. everything that happens outside the CPU, CIAs, SID and VIC-II. That means every single memory access has significant overhead because it is essentially simulated from a transitor netlist similar to how visual6502.org simulates the CPU (which even in an optimized C simulation like https://github.com/mist64/perfect6502 is still about 30x slower than a real 6502 at 1 MHz).

PS: Let's do some simple napkin math: the Antirez simulator is 3x slower than a real C64, 100x faster would mean around 30x faster than a real C64, e.g. like a C64 clocked at 30 MHz. That means on a 3 GHz host CPU you would have 100 host clock cycles to emulate a (30Mhz) C64 system clock cycle, and that's for 5 complex chips (CPU + 2xCIA + VICII + SID), so around 20 host clock cycles per emulated cycle for each chip, and that doesn't include address decoding and memory access (which is the expensive part in Antirez's simulator). And moving the chip emulators into threads won't help you much here, because you'd need to synchronise at least once per emulated clock cycle (maybe even twice per emulated clock cycle because of the C64's phi-1 and phi-2 'half cycles').

Good luck with that :) (e.g. squeezing a complete VIC-II clock cycle which renders 8 pixels and may have up to 8 active sprites into 20 host system clock cycles, even with perfect pipelining, SIMD, and no cache misses this isn't a lot of host CPU instructions compared to what a VIC-II needs to do each cycle).

(disclaimer: I'm the author of the pin-based chip emulators used in this experiment)

I think this is a rare “useful” abuse of font ligatures, and in that respect I think this is a cool project, but it's also a terrible idea that nobody should actually use, because Japanese kanji readings are famously ambiguous and context-dependent. The modal character has two different readings without counting those used in personal names, nor contextual slight changes to pronunciation. Some characters have six different readings, and that's again without counting those used in personal names. There's no way a font can do a good job at it unless you can somehow stuff an entire LLM in there.

For context, it's a hard enough problem that Google Translate's own romanised pronunciation guides are frequently wrong, or disagree with their text-to-speech.

I have noticed the same on my project.

The agent helps a lot to fix bugs. Just point the agents at the bug reports. They do the reproduction, debugging and fixes. We "just" have to review. (Sometimes ask more details and tell we'd prefer another approach.) This makes bug fixing much faster and automated.

And at the same time, it seems we get less bug reports to our library. I think user just use their agents to code, and agents don't report bugs, they just work it around somehow.

So overall, the open bug count is decreasing. Which somehow feels good.

On the other hand, we get a lot of PR made by other's people agents, which can be a bit overwhelming when it's just a huge code drop. We can't easily review that.

If I let a "self-driving" car run and it smashes into things its my fault.

If OpenAI does it, it's gone rogue and it is nobody's fault.

They need to be sued, the laws are already there.

Your prompt, your responsibility.

What's really striking is the current administration performed one of the most brutal cost cutting programs in recent memory. And despite the massive amount of disruption caused (and to be caused) there is almost nothing to show for it financially. Just pennies in the ashtray of a Lamborghini.

Obviously the people behind it were not serious, and did nothing to actually address government waste or reform the welfare system. But the result is the same - the idea of cost cutting as a means of addressing the deficit is effectively dead. There is no dry powder on the "slash budgets" side of the ledger. All that's left is raising taxes. (Or do nothing until there's an actual hyperinflation event).

As the zoomers say these days "Unstoppable scroll vs unskippable title"

I have heard from someone who works at Anthropic that they will have postmortems that sometimes devolve into "my claude says X" vs "well my claude says Y". Maybe this is the future of software engineering, but it seems pretty fucking sad.

so i did some apples to bananas comparison.

i let 4 different agents review the same dirty code base (200+ files changed, c++, game engine):

- deepseek v4.1 flash via deepseek API: took 20m to find one p1 issue that was relevant and a p2 issue that was irrelevant

- glm 5.3 flash NVFP4 via TP2 DGX Sparks: took 1h 5m to find a p3 documentation issue

- gpt 5.6 sol: took 5m30s for one p1 and one p2 issue. both relevant

- gpt 6.1 sol: took 4m30s for two p1 and one p2 issue. both relevant.

Worth to mention: none of those findings overlapped.

I let deepseek v4.1 flash and gpt 6.1 sol both plan fixes for the summary of the findings. in this particular scenario: gpt 6.1 sol planned to increase a buffer size from 128 to 256 to fit the 191 elements that blew past a test scenario. it included some headroom. deepseek v4.1 flash on the other hand decided to make it exactly 191 to satisfy the test, for example.

The TPS decode on deepseek is impressive. I never had a second / feeling of being left aone and though "damn this thing is working". Despite all my efforts, it took ~4x longer in a review, found less and planned worse than GPT 6.1 Sol. Wall clock time. I don't have token counts for comparison. If I were to grade those models from 1-10 (10 = top tier GG), I'd give deepseek a solid 6.5 and 6.1 sol a 7.5-8.0 maybe.

> In the town of Abu Falah, near Ramallah, at least 80 settlers, some of whom were carrying firearms or clubs, gathered to stop Palestinians from harvesting their olives, using slingshots to hurl rocks at the farmers in view of Israeli military vehicles. When some of the Palestinians then picked up rocks to defend themselves, the Israeli military fired teargas in their direction.

This has become beyond ridiculous. I have personally gone from full support for Israel to now not even supporting Israel's right to exist. And I can't be the only one. I feel sorry for the (clearly increasingly rare) "normal" people of Israel but they will soon need to move out of Israel to not be bundled in with the crazies.

As an aside, I do not like unnecessarily misleading graphs. Own your strengths and weaknesses.

In the "Results" section, they choose to compare Ultramarine only to models that it's better than on every metric. They ignore models like Fable 5.1, GPT-6 Astra, Claude Opus 5.5, etc.

Then, in the "Score against cost per task" graph, which is focused on price, suddenly all of those models that perform better than Ultramarine come roaring back because they are also more expensive.

Just list yourselves as the second best sometimes. Your thing is impressive. Don't let the idiot marketing guy hide that behind stupid tricks.

>“There’s not anything like this in the world,”

Must've never driven through the Alps I guess. There's something like this every couple km