Our experiments demonstrate that tapered-precision arithmetic consistently outperforms IEEE 754 floating-point numbers in all tested scenarios. Among the tapered-precision formats, takums exhibited exceptional performance, outperforming bfloat16 in every case. While occasionally marginally less accurate than posits, takums delivered comparable results overall and demonstrated superior numerical stability. Notably, we successfully introduced the application of 8-bit posits and takums in mixed-precision iterative refinement, marking a possibly significant milestone in numerical computing. Additionally, GMRES exhibited particular benefit from tapered-precision formats, with takums delivering outstanding results, surpassing posits in all cases.
Wouldn't a better formulation be: "Give me the shortest route with a cost of 5 or lower with the minimum fire tiles in the path necessary" Since your formulation doesn't care about the amount of fire tiles in case there is no other solution?
In my opinion, you should certainly go for 1.0, unless you are planning some breaking changes at this point - which I guess you are not, after reaching beta. My general deduction is that tools with a few bugs at an early 1.0 has a better chance of succeeding than tools that stay too long in beta. It's important to ensure that interest from users, distro packagers and commercial adopters are kept alive while waiting on 1.0. I'm still waiting for pijul to be in the distro repos, though the crates version work satisfactorily. This may turn away many other potential users. Alphas and betas should be time-limited, not bugs-limited. Bugs are what patch versions in semantic versioning are for. Remember: 1.0 is a magic number - it increases the exposure of your tool and potential contributors by a large factor. By your standards, git would have stayed in pre-1.0 for a decade and would not have become this popular.
On a personal level, I use pijul for managing my dotfiles. I haven't encountered any bugs yet. The only feature I miss from git is the availability of tools like syntax highlighters for diff and record and UIs like tig, gitui and magit. But again, those are just features you could add as minor versions. That aside, pijul feels like magic. A handful of pijul commands cover most of the use-cases of git. After using git for so long, I always wonder if I'm using pijul the wrong way, even though I get the expected result. I'm only familiar with svn, git, bzr and hg. Perhaps Darcs would have given a better idea about what pijul really is.
I was thinking of doing a synchronous release of 1.0 along with the new Nest: the current one has a cool architecture (replicated over 3 datacenters, using Pijul repos as CRDTs), but doesn't scale well, I'm working on a serverless version. The hard bit is convincing Pijul that it's talking two a real Pijul repo, whereas it's actually talking to a serverless cloud (that's already solved btw).
EDIT: I also don't think there's a "right" way to use Pijul. The magic of algebra means that you can't possibly break the associativity and commutativity properties of Pijul patches, so you can do whatever you want, I'm sure it's fine.
Thank you for the reply! I'm eagerly awaiting the 1.0 release. Good luck to you regarding the new nest and well as the new venture!
> I also don't think there's a "right" way to use Pijul. The magic of algebra means that you can't possibly break the associativity and commutativity properties of Pijul patches, so you can do whatever you want, I'm sure it's fine.
There is a certain time to 'internalize' the concepts of a tool like git or pijul. It took me years with Git and I'm only starting to be confident about my git skills recently. I haven't had that much time with pijul and I haven't absorbed its concepts yet. Despite that, pijul feels like a concept that I can learn much faster. That causes this 'sunken cost' fear. If it was so easy to learn a VC tool, why did I take so much time with Git? That's what I meant with that comment.
I'm a big fan of pijul. I even have a plan in a future project to manage editing history using pijul concepts (or library).
Everything needed to go from Q > 1 to actual electric power is nonexistant, not even being worked on, and far beyond the resources of these companies. Nobody knows how to produce the tritium needed to operate, or to extract it at ppb concentration if they did. Nobody knows how to build structural elements that will not fall apart after a few months' neutron blast. It has been decades since either was seriously looked at, with no promising ideas forthcoming.
That hasn't mattered much thus far because Q is so far off, and investors pour money in regardless.
This is not the will of the people in the Netherlands. We had a referendum a few years ago about the laws for spying (sleepwet). Their was a clear vote for not allowing bulk spying. The rules that currently are broken down where made in reaction to that referendum.
I've looked up the figures again: 51.5% turnout, of which 49.4% was against, 46.5% in favor. That's not a clear vote, and the political climate hasn't really become more favorable since.
That's a pretty clear "hey government, how about you stop trying to push things through and do some more asking around whether people actually want this" if you'd ask me.
Of course, it's easier to just keep trying and altering your methods until people give in from exhaustion / negligence, rather than actually figuring out whether the people want it.
I hate the youtube search these days. You get back a very limited list off content you serached for and a lot off videos that are "related" to your search.
Have you tried searching for videos using Google, DDG, or Bing? I find it's better than YT's search. There is a browser plugin named "Unhook" that removes "related" search results.
Except votes for parties who get less than 5% (if running alone) or 8% of votes (if a Coalition) do not count. So that 40% becomes 52% and suddenly the only thing you can do as a minority party is yell.
We haven't legalized the production of drugs in the Netherlands, so our 'gedoogbeleid' does not help against organized crime. It does help on the health side of the debate though.
