I’m surprised there’s such a big turnover for packages. Looks like the total number of packages is on the order of 100k, so 20k updated packages sounds about right to me but 20k added/removed seems high, especially since nixpkgs was already pretty much the most complete set of packages out there. I guess there were a lot renames that their statistics did not catch? The last two releases had around 7–8k new packages.
Perhaps of interest: I came across this retrospective of the author of that site and how he moved it to static hosting a couple of years ago [1].
I would also say that this library covers more or less the “lower half” of solo ball juggling in terms of difficulty. With lower ball counts (say ≤ 4), there are a lot of these patterns that have complex arm movements and can be difficult to explain with words, so having such a listing with animations and step-by-step instructions is very valuable. Starting with 4 balls, there’s less and less time for moving your arms around and it is more about the sequence of heights of the throws, which are well described with just their numeric “siteswap” pattern and you can learn them just from knowing the number sequence. The site has only the most basic of those (e.g. 534) and even very common 4-ball (7531, 633) patterns are missing with hardly anything beyond that.
If you are just creating a simple document with default styling, the main advantage you get from Typst is near-instant compilation speed. Pandoc to HTML is similar though, but if you’re generating PDFs with LaTeX the compilation delays can be pretty annoying.
If you are creating more complex documents, the advantages become more pronounced. Styling in Pandoc means modifying templates, at which point you’re just writing LaTeX, and styling in Typst is much nicer than in LaTeX. You can also hit the limits of Pandoc templates quite easily, at which point you have to write Lua filters. I have found those to be quite cumbersome, and now your document logic is spread out over the Markdown source file, the LaTeX template, and the Lua filters. In Typst you can have a single file with your whole document in a clean modern format, and you can decide for yourself how much you want to separate content and presentation.
It’s also what Thunderbird is using to build their paid email hosting. Seems like a very ambitious project mostly done by a single person – impressive!
The issue I have with this is that it deletes the whole message, not just the media. In WhatsApp, you can delete media from the images/video folders and the messages remain in the conversation, they even still have the blurry preview iirc. In Signal, you end up with gaps in your history instead.
From what I can gather, they are not yet incorporated and they are working through organizational questions in [1] and the issues thereof. The `ACCOUNTS.md` file there gives an idea about the main people behind the project and the donation page on Open Collective [2] also documents team members and how they spend those donations.
Technically the big O notation denotes an upper bound, i.e. it doesn’t mean “grows as fast as” but “grows at most as fast as”. This means that any algorithm that’s O(n²) is also O(n³) and O(n⁴) etc., but we usually try to give the smallest power since that’s the most useful information. The letter used for “grows as fast as” is big Theta: https://en.wikipedia.org/wiki/Big_O_notation#Use_in_computer...
That’s exciting! Is there a summary of what’s new in 1.0 somewhere? Usually the version announcements contain an overview of the most important changes, but this one is more of a retrospective and general overview apparently.
You could still move e.g. /nix to the BTRFS disks and only keep /boot on the thumb drive, no? That’s more or less how I have NixOS set up with ZFS. But your solution is of course much fancier :D
I’d love for Purism to succeed and they appear to be doing a lot of valuable work in the software space, but I can’t help feeling like that 75M self-assessed valuation is a bit on the optimistic side…
I’m not sure about these stable packages, but for the nightly packages they introduced recently they explicitly mention on [1] that you can keep browsing:
> Following community discussions, we have updated the post to highlight that Firefox can continue browsing after an APT upgrade, allowing people to restart at their convenience.
From what I can tell, the “Ubuntu Mozilla Team” does not consist of people from Mozilla but rather just people packaging Mozilla software for Ubuntu. The latest packages in that PPA have been uploaded by Rico Tzschichholz, who does not appear to be affiliated with Mozilla.
I’m continuously impressed by the care and effort that goes into the developer experience of Elixir. Those improvements of the (already pretty great) documentation and the error messages look like they will make a real difference.
The way I see it, the problem is the C in CSS. If you rely on `.author-bio > h2`, then every time you make a change to `.author-bio`, you have to consider all the contexts in which this class may be used globally in your site, and if you change the styles for `h2` even more so. Sure you can say you encapsulate every context with a globally unique id, but since you’re probably already using some kind of templating system for you HTML, it’s easier to specify the styles there. It’s basically the difference between having variables scoped global vs. local by default.
I would say there is one more problem it can solve, though less related to sustainability: Building up a modular system may make it easier to more offer customers a device close to their ideal configuration with a limited number of SKUs.
One distinction that I had missed for a long time is that the en-dash is used instead of the hyphen to connect words in the “and” sense, such as “read–eval–print loop” or “Myers–Briggs personality type”. I find that the en-dash makes it a bit more clear that the words are sort of “on equal footing” and it’s not one word modifying the other one.
I wish it was possible to add FIDO keys to an account without having physical access to the key. Without this, it is hard to balance the convenience of adding your keys to new accounts and the risk of losing all your keys. Ideally, I’d want to keep one key in a safe location far away and just have some public key data that I can upload to new accounts. Does anyone know why FIDO doesn’t work this way? Is it simply to make it harder to lock yourself out of an account?
I’m surprised there’s such a big turnover for packages. Looks like the total number of packages is on the order of 100k, so 20k updated packages sounds about right to me but 20k added/removed seems high, especially since nixpkgs was already pretty much the most complete set of packages out there. I guess there were a lot renames that their statistics did not catch? The last two releases had around 7–8k new packages.