seems loved languages such as python & ruby (ZJIT | TruffuleRuby) have been getting a lot performance improvements lately. of course JS with v8 kickstarted this - followed by PHP.
so for majority of us folks use what you love - the performance will come.
As someone who was a hardcore python fanboy for a long time, no, no it won't. There are classes of things that you can only reasonably do in a language like rust, or where go/kotlin will save you a crazy amount of pain. Python is fine for orchestration and prototyping, but if it's the only arrow you have in your quiver you're in trouble.
I know sometimes performance doesn’t matter, and python is certainly useful, but it’s not fast. It can be fast enough and they’ve put a lot of effort into making fast libraries (called in c).
When doing bioinformatics we had someone update/rewrite a tool in java and it was so much faster. Went from a couple days to some like 4 hours of runtime.
Python certainly can be used in production (my experience maintaining some web applications in Java would make me reach for python/php/ruby to create a web backend speed be dammed). Python has some great libraries.
I like the convenience of the cloud. but don't know whether its due to declining literacy rates / awareness etc. the cloud is nice and e.g google storage, iCloud but now with fast microsd's you can buy 1TB for $100. have a few copies then boom, you own your own data. but now phones don't allow you to have microsd's so here we are.
likewise things like email etc instead of all of us being on gmail we could have community email servers etc.
I use Dropbox, but with an encryption overlay that also integrates into the iOS Files app for ease of use on mobile. So it’s possible to use cloud storage and still keep your data private.
John Carmack was well-known for keeping his development notes and other thoughts in a .plan file[2][3]
Aside: it saddens me a little how the simple idea of copying files onto a server is so many steps removed from modern "web framework" behemoths. I think some people genuinely don't know about the simpler older ways, and the pros and cons between them.
I use org mode files, a small build script and git.
The build system just invokes emacs and compiles org documents to HTML, and installs them in /var/www/${site}. I have a git update hook on the server that invokes the build script when I push updates.
Originally I did just rsync over HTML files though, but I like the new setup a lot.
quite sad, the space had inspirational stories. & IndieHackers brought those stories to the masses to inspire.
but later 'fast money / influencers' entered the space. It has become a mini ponzi sell a starter pack / template to wannabe indiehackers, sell a course to wannabe indiehackers.
For solo founders I guess if you wanna get all meat no bones stick to MicroConf
worrying about H1B is looking at the wrong metric. Most companies are past H1B - the game is now offshore stuff.
Look at the data: e.g a company like Rippling where a good portion of Software engineering jobs are in India. then Senior / Staff Roles are in the US. Most scale ups startups have similar dynamics. It could be India or Eastern Europe.
Part of it is that most of the libraries that came before we’re tightly coupled to a particular framework that itself went out of fashion, like Passport and Express, which is a problem because frameworks themselves have been moving in and out of fashion very rapidly; or are coupled with service offerings from vendors, like Auth0.
Auth.js is actually one of the first attempts that tries to be framework and vendor agnostic while still including a good deal of the batteries you need to make a full authentication system, which they only recently did, as they were originally tied to next JS like every other library in the graveyard of authentication libraries.
If you just want to specifically do an OAuth handshake or salt and hash a password or produce a JWT, those libraries are all rock solid. But a full batteries included framework and vendor agnostic solution hasn’t really existed until recently.
Same. I've personally never had issues with any auth packages, granted I've never used auth0. Personally, they all seem quite similar, especially in the react world.
Anything that can help me utilize oauth standards is fine to me.
It’s not that auth is unsolved in other languages/frameworks, but it’s often way too complex or configuration-heavy. If adding passkey support to my app is going to take 2 hours, that’s two hours I’m spending away from building my core product. For smaller projects, that’s not time that I could afford.
For example, if I want to add passkeys to my .NET CORE app, this is the guide Microsoft provides:
For some projects, the flexibility of other solutions might be needed. But for ease-of-use and development speed, better-auth has been a clear winner for me.
Excuse me, incoming contrarian!
learn.microsoft, is for learning about the concepts as well as the practical applications. Also for user facing security, wouldn't you want all the knowledge available to you? Much easier to find the foot guns in these kinds of situations.
Rust is a good language for performant computing in general, but especially for data projects because there are so many great OSS data libraries like DataFusion and Arrow.
SedonaDB currently supports SQL, Python, R, and Rust APIs. We can support APIs for other languages in the future. That's another nice part about Rust. There are lots of libraries to expose other language bindings to Rust projects.
so for majority of us folks use what you love - the performance will come.
reply