naranha

"Open - for some definition of open" AI

Prepare to open your wallet each time you use it.

For me something to stand out in the current day and age. THE "killer" feature.

It has some nice things but for everything in PHP, there other languages that do it a lot better. And I don't think CGI is a killer feature in 2022, it doesn't matter anymore than any other implementation detail, and causes more problems than it solves.

Yeah I wish the focus was more on backwards compatibility at this point. Most PHP codebases are very old.

> Java comes to mind.

Java used to be like that, but the migration to the module system has been very rocky for me personally when revisiting older projects.

If the switch proves anything it's that you don't have to pay $1600 for a GPU to enjoy amazing games. Or the other way round, if you pay $1600 for a gpu it does not mean that you'll be able to play any good games.

Maven is a pretty horrible dependency manager. I would take NPM over it any day, because it is so much simpler. There is much less that can go wrong, while with maven you're fighting with the tooling all the time. pom.xml files are much larger than package.json files. While I can read and understand a package.json no problemo in a couple of seconds, with pom.xml files I'm completely lost for hours on end, even when I wrote them myself a while ago.

I think TSX/JSX is one of the top reasons to use React. It is so tightly integrated with Typescript and the IDE/VSCode that it allows you to type check your markup at every step. That was one of the biggest things I missed in Vue (2). But I think these days Vue also supports JSX (?).

Other reasons are that you have a lot of control over what gets re-rendered when using memoization, with the huge but that there a lot of foot guns. As well as it is easy to write terribly performing code in react, while it is really hard to write fast/good code.

> Something like what you describe might be more efficient on certain queries but only because it’s giving up a lot of flexibility - consider what happens if you add any filtering or grouping changes and those partial sums can no longer be used.

If you're doing a query that is not backed by an index, it's not scalable, so you could also do it in Excel.

> This preference for general systems rather than optimizing for a single query shows through in the tools used for this kind of task: materialized views and indexes using expressions.

Materialized views totally suck (at least in postgresql), because you cannot update them incrementally (yet). A better option most of the time are triggers and aggregate columns, but it requires a lot of micromanagement.

> It sounds like you don’t want a general purpose database

I don't think I want a special feature, map-reduce indexes seem pretty general "mathematically" sound thing.

> Another way of looking at this is to ask why a field with billions in R&D hasn’t already done it, and especially why various big data tools have been significantly less impactful than promised.

Well that's what's exiting about research, that some things haven't been done yet. Even though they may seem fairly obvious.

But even SUM in AWS RDS Aurora is not optimal in terms of performance. You could do much better by having partial SUMs in the index btree nodes, that way you could get the sum of billions of records without processing any row individually. CouchDB supports this (even though it's by far not the most efficient database). I wish Postgres and others provided access to that sort of low level map-reduce index functionality. Transferring large amounts of data in the response is solvable by using shared memory, so I'm not concerned about that. And a SQL like language could be provided on top of that. If a low level interface was standardized you could even have SQL like languages that are portable between DBs.

My problem with SQL is that it's 99% Excel and 1% A database query language. And the boundaries between are often not well defined. Why even have a SUM function, when it is not accelerated by an index, that makes using it automatically non-scalable. I think there should be a core language, perhaps similar to SQL perhaps not, as an interface to pure DB functionality, all the other 99% could be done in Excel and while it's nice to have, it should be more clearly separated.

Sounds like what DDR5 is doing too, errors are corrected automatically, but not necessarily communicated (?)/

Do you think Raspberries without ECC RAM are fine to use for a Ceph storage cluster? I did some research yesterday on the same topic, many say ECC Ram is essential for Ceph (and ZFS too). But I'm not sure what to believe, sure data could get corrupted in RAM before being written to the cluster, but how likely is that?

I mean in 2022, there are so many options in Linux that may not be perfect, but are good enough. Be it KDE,Gnome,Wayland,Xorg,systemd,initd,bash,zsh they are fine to use. Yeah you'll hit a corner case here or there, but most people will find a solution that fits their workflow. Windows on the other hand will always be the Gamer or enduser-that-don't-care OS, and that's fine too. I'm incredibly thankful that Linux exists, because I personally gained so much from it, but I understand as well when people don't care about it. Let them use Windows and deal with their problems (but don't expect me to fix them for you, thanks ;)).

Just want to point out that logically there is a third option: Renewables + storage + other energy source (nuclear). I wonder why discussions around energy always end up in false dichotomies like nuclear XOR renewables, when the most logical solution is to use every option we have to solve climate change.

When microservices were new they brought a whole slew of new technologies and architecture patterns with them: SPAs, JWT authentication, micro frameworks, REST/Graphql, containerization. Things that solved problems people were having with the previous monolithic approach, all above the lack of maintainability and composability. So I see the term microservice today not as something that's measured in lines of code, but above all by embracing the new technology landscape that came from it.

Luckily here in France there are 8 static /64 prefixes included in a standard DSL or Fiber offering. Just luck i guess...

Hmm yes I have several static prefixes, so that helps. I'm in the EU though...

Just a $40 openwrt router basically. It uses Jool.mx for NAT64. And I simply use Google or Cloudflare for DNS64.

I think Android includes it too, because some services bind to clat-like ipv4s. But I have not investigated this further so far.

I'm running my home network IPv6-only since some time and it works fine thanks to DNS64/NAT64. I think once more ISPs start offering DNS64/NAT64 internally the transition will be quite unnoticeable for endusers.

Software that still does not work because it uses hardcoded ipv4 addresses or sockets: Steam, WoW (probably many other Games too), node/npm (before version 17), but for the most part it works! The offenders can also mostly be fixed using clatd.

Works great from my experience. What's missing and why I'm not using it atm is DNS64 and TLS support.

I'm glad that today there are more alternatives and it is not only .NET vs Java for the enterprise segment. Since the 2010s NodeJS, Go and Rust emerged and the world would not end for open source if .NET or Java went completely closed and many already moved on.

> Brackets are queried when rendering the viewport and thus querying them has to be really fast.

Then by querying them only when the document changes could make it another 10,000x faster ;)

Apparently it used at facebook a lot too.

https://lwn.net/Articles/824855/

What is kind of neat for your case maybe is that you can run btrfs scrub even on raid0 or a single drive. The scrub command reads the whole filesystem and verifies checksums, so you can be sure that the data is readable and correct.

Disclaimer: I only tried it on a single disk.

PostgreSQL has a lot of really great features, but some things are seriously underdeveloped as well, mostly because missing man power. You cannot reorder columns of a table easily, it does not support compression of connections (yet), XID-wraparound as others have already said.

I would also personally love map-reduce indexes (for example for efficiently calculating sums or other aggregate functions...)

But you can also send encrypted E-mail using GMail with Thunderbird...

Replace "Swiss" with "US" and Protonmail with GMail and the sentence remains equally true. So I guess the question remains, what does Protonmail offer in terms of privacy that is better than GMail or Outlook, given that USA and Swiss are both adhere to democratic standards.

Yes, this seems to me like mostly FUD, because there is no line of reasoning to be found (I read it on outline.com). It says it takes too long to innovate, without any concrete evidence, afaik there are many innovative reactor designs waiting to be constructed. Maybe not next-gen, but better than past ones that need to be shutoff at some point. Also the old design PWRs still work well and can be constructed today within a few years, so definitely worth pursuing.

Economically more viable than burning stuff that literally comes out of the ground and there is near endless supply but destroys the climate? Definitely not, but neither are renewables. The economy won't safe the climate.

Arguments against nuclear would probably be fuel availability/harvesting and safety.

I think docker/podman with firewalld behaves the same way. I was astonished when I found it out. I think you can however bind to localhost exclusively by using the ip in the argument:

docker run -p 127.0.0.1:5555:5555 ....