Still, there's a part of me that sort of wishes there was an attribute like

#[just_use_a_uuid_for_this_symbol_name_pls_i_expect_it_to_be_large]

IIRC a lot of it was because there's an implicit cap in some MSVC linkers. Also seems to depend on just how clever the inliner gets - sometimes the long symbols just vanish before rubbish linkers need to choke on them!

My "this seems possible without negative side effects" t-shirt is producing a lot of questions, etc.

I believe they call this 'cooked'

Stares nervously at the many 'linking failed because symbol is too long' issues in the chumsky repository

Idk, a gibberish machine can't be any worse at the job than he is

Surely there's a mechanism by which ministers can reject the ruling on the basis that it's entirely contradictory on its own terms? The supreme court might as well have ruled that pi is 3.

The idea would be to deprecated the old methods in future additions. That way you get the benefits in future code without causing problems for maintainers of existing code.

The price of housing is driven mostly by income because there's no consumer surplus in rent, so gains mostly just get piled into property value inflation. Makes those on the bottom end of the income scale more competitive, but due to fixed supply, more competition only means more misery for everyone

'We recommend this new method going forward' is 10x less annoying churn-wise than the sort of arbitrary changes clippy makes to its default lint set every few weeks.

Agreed that it's broadly a non-issue, but it doesn't feel like it needs to cause pain for... well, anybody

Would there really be that much churn? Deprecate the old method, add a new one. An edition could easily hide the old method from new code. Also pretty trivial to have cargo fix handle it. This seems like exactly the sort of change that Rust's laser focus on backcompat should make easy.

Highly unlikely to happen, unfortunately. There's a lot of money in keeping the ball rolling, so the protocol will progress to post-quantum cryptography. That aside... we're much further from viable quantum computation than the technology brothers would have us believe.

LLMs give people with delusional belief systems a way to organise those beliefs. The model fills in gaps, supplies connective tissue, and wraps the madness in formal language.
It doesn't make the claims more credible to us, it just tidies them up.

Do you think there's a logical mechanism for separating them? I can't imagine how one might go about designing such a system. And, of course, it would means that the bulk of society misses out on any potential gains too.

The fact that risk management still assumes that the performance of individual stocks is largely independent across a large portfolio just provides an avenue for these silicon valley creeps to socialise risk.

I can accept a small number of crackpots engaging in that sort of behaviour, but the erasure of risk through many layers of investment repackaging - and how it exposes the rest of us to that risk society-wide - is truly quite scary. The stock market was supposed to dampen shocks, not magnify them.

When people start talking about 'hype curves' and 'plateaus of stability' it's usually because they don't actually understand what the fuck they're talking about and they're just jamming the thing into the same square peg hole they think applies to all of reality because they saw a made up graph