mcint

I don't have the spare cycles to substantively engage. These posts would make for really good discussion seeds in the linux infrastructure meetup I run. Comprehensive, but still approachable answers to what it takes to solve X well, here font rendering.

It means, at worst, don't go outside or be in bright sun-lit spaces, but also, we can shine a flashlight on where the cancer is to kill it. Even if it weren't very selective (which could be adjusted with other targeting molecules), light-trigged theraputics are easy to understand the propagation of.

China also builds nuclear reactors, and we can't fall behind them. I cannot abide an SMR gap.

https://youtubetranscript.com/?v=ouF-H35atOY (same video, I'd watched previous to seeing this hn post)

  other model details the authors note
  that most prior State space models use
  complex numbers in their state but it
  has been empirically observed that
  completely real valued State space
  models seem to work fine and possibly
  even better in some settings so they use
  real values as the default which work
  well for all but one of their tasks next

  it succeeds on test sequence lengths of
  up to a million tokens which is 4,000
  times longer than it saw during training
  while none of the other methods compared
  to generalize to Beyond twice their
  training length

https://youtubetranscript.com/?v=ouF-H35atOY (same video, I'd watched previous to seeing this hn post)

  other model details the authors note
  that most prior State space models use
  complex numbers in their state but it
  has been empirically observed that
  completely real valued State space
  models seem to work fine and possibly
  even better in some settings so they use
  real values as the default which work
  well for all but one of their tasks next

  it succeeds on test sequence lengths of
  up to a million tokens which is 4,000
  times longer than it saw during training
  while none of the other methods compared
  to generalize to Beyond twice their
  training length

All manually user-generated by people contributing <= 3 names through https://www.gnoosic.com/, then voting on other names suggested in the same neighborhood of the graph.

It would be cool data to get access to. But it's a nice singleton in the mean time. Like Netflix's challenge dataset, it would be interesting to have another layer of associations to compare with more formal genre lists. I wonder what kind of data Spotify releases.

from https://dortania.github.io/OpenCore-Legacy-Patcher/MODELS.ht...

I thought my 2015 MBP wasn't able to upgrade. Good to know it's still supported.

Assuming you mean password manager services. If they can get a copy of the vaults at all, in a well-designed system they have to attack each one individually.

They also serve business use cases for sharing, and revoking access, to other services.

> So, why don't people use local password managers?

Because synchronizing between devices is an important factor in usability, even just for an individual, not even considering groups or businesses that would share access.

Local password manager vaults can also be stolen by malware, would result in the same catastrophic loss you suggest.

> Just a txt file encrypted with "master password" should be pretty damning to break into. And the reward for breaking in would be password for 1 person. (compared to 100k businesses).

Password managers do just "do the thing you'd want them to do", and because they have a well-defined use case, they can support extensive discussion about threat models, and can easily and coherently support lots of people benefiting from the best research and security, without each having to individually roll a solution.

A hosted service without persistent compromise is likely less vulnerable to an old copy of a vault, and an accidentally disclosed master password causing catastrophic failure.

Lastpass has some issues that look egregious in retrospect. The early exceedingly low-strength vaults that were never upgraded (only possible on login).

1password publishes research and pushes forward authentication management beyond password, e.g. passkeys, in a credibly cross-ecosystem way.

Not a fan of the

  html { font-size: .78125vw; }

I made more progress once I accepted (possibly) non-uniqueness and worked on finding some valid assignment, as opposed to a sole valid assignment.

The last dozen clues were increasingly interesting to solve, the intersecting constraints.

https://www.youtubetranscript.com/?v=dkpDjd2nHgo&pp=ygUWbWwg... though only /rulia/ matches, only once. Not a great transcript from YouTube it seems.

Thanks checkmeyer for devoting some time to this resource, I really appreciate having hn history available at a keypress!

  &type=story&type=comment

  &type=all

https://circuitlaunch.com/ (working hard to brand away from /makerspace/ or /hackerspace/, but serving a similar audience and purpose, in a professional/entrepreneur/business-building direction).

There are many! Also, more importantly, other institutions have taken up ideas from them and offer tools and services to address the same needs. Libraries and colleges in many communities host makerspace tools, tool libraries

The maker movement has taken off, Etsy, home crafting, and sharing and selling projects at all varient of levels of polish,

Finding like-minded individuals, for all manner of interests, is possible through meetups.

I suppose, they've unbundled. I would say: look closer, they're maybe more present than ever, but in parts, and adapted in ways that best complement the psychological demographics of the places they serve.

[1]: https://wiki.hackerspaces.org/List_of_Hacker_Spaces [2]: https://noisebridge.net/

I did use the "all" search, I made search engine shortcuts for all "hn", stories "hns", and comments "hnc" in my chromium* search engine settings. The all search now returns nothing. I like to cast a wide net in search results. I don't think I'll search for many of these categories in most queries, if pressed, but I also won't discover them. May motivate me to index hn myself as some of my friends have.

I appreciate surfacing the other formal and informal categories of posts in the hacker news repertoire, and it's fantastic to be able to filter strongly by category, but I don't like that I will now have to always pre-select, pre-commit, to the singular category I want to find matches in. I suspect some halfway solution, like match-counts of other categories, would be nearly comparably expensive to returning results for all, in terms of query cost and maintenance or complexity burden.

I'm also seeing text styling issues in the menus in settings: white text on white background, in both default and experimental skins, and light and dark modes of experimental skin. I checked in incognito to be sure it's not my extensions affecting display, nor Brave resource blocking.

I'll echo the value of ijq, https://sr.ht/~gpanders/ijq/ The live updating makes for less to remember and visualize. Though there is some value in having intermediate history recorded, for the second and later questions you ask of the data. (And supports commenting out for error diagnosis without deleting half the expression, `# ...`, hash and space required.)

I hadn't known of `debug`, quite convenient. I've often used multiple outputs at the tail, for example `... | keys, .` to remind myself of the structure as I explore down. Also `gron` and `fzf --multi`.

Logically, reaching for a justification, an explanation, sure, it's reasonable that the MSB flips halfway around, but again it differs from the analog clock reading conventions.

v2 (+/v2.html) which he posted on another thread, does make it clearer

https://retr0.id/stuff/2038/v2.html

Other thread of his blog domain

https://www.da.vidbuchanan.co.uk/blog/unix-clock.html

https://news.ycombinator.com/item?id=37492371

The color separation is also a bit large, mainly 2-3 and 3-4, while 1-2 are similar but distinct.