Recent comments in /f/technology

flabberghaster wrote

100%. They bill it as "you don't have to pay a guy to run your on prem hardware for you" but you absolutely do not save money once you've scaled to the point you need a few machines.

And by that point, your software is probably so tied in to proprietary Amazon stuff that they've got you in a vice.

It's the same model as Oracle: oh we're so easy to use and friendly, and we have all these great Oracle specific features, go ahead and use them! 😊 but then they have you locked in and they can just start squeezing.

3

twovests OP wrote

Going "your personal blog" mode: I don't like the bespoke names like "Route 53", but at least a lot of AWS makes sense? There are very few things I've come across (as someone toes-deep into AWS) that seem like a horrible engineering mistake made at Amazon.

If I were to build an intranet for the internet and divide everything into tiny little separately-billable services, I'd do it a lot like this.

4

twovests OP wrote

everyone laughed at me when i said i had an idea for interactive visualization tools which work up to 5 spatial dimensions :(

that would have been great for manifold learning, because usually you only work with 2 or 3 spaces.

if the true shape of something is 10 dimensional or less, you can explore that by umap'ing it to 5 dimensions or less

alas

2

Jenheadjen wrote (edited )

Reply to comment by neku in The Juicero of Bikes by voxpoplar

Very different kind of product but i think the thing that most recaptured the magic of Juicero was Quibi imo. In terms of like, massively overfunded thing that everyone immediately recognized as a stupid idea and was widely mocked throughout its short existence.

2

neku wrote

i think this is quite juicero but i think the og juicero is inimitable for their sheer chutzpah and the fall from grace afterwards. the way they hyped up the joy of cold pressed juice with the Juicero Juice Press, the $800 price tag only to find that the bags could be squeezed by hand, the way it felt like they were the focus of the entire internet that week... juicero was once in a lifetime. this is an overambitious kickstarter. juicero had vc backing

5

neku wrote

Reply to HarmonyOS NEXT by twovests

i think huawei is an interesting company because they're basically building a technological ecosystem independent of western interests and intellectual property. plus they make some really nice stuff. i got a huawei nova 5t phone some time in 2020? and it's still running great in 2025, and their screens are beautiful. only trouble is that they don't always sell in the west

2

flabberghaster wrote

this is assuming the only thing the server is running is linux apache mysql and php; but you'd just write your PHP code and create a package like an RPM or what have you, that deploys it to the right place, and your configs for the rest of things would also either be their own packages, or managed by some script or puppet.

It all depends i guess. I suppose it probably is much easier to manage if you just compose some images, than if you say "you gotta configure the machine" because if you can abstract away much of that stuff into containers then you're not stuck on one distro of linux, so i get why people use docker (which I call dorker btw).

3

twovests OP wrote

a bandaid for having too many dependencies or a workaround for people who don't want to make their software easy to deploy

I'd be curious to ask this; what could someone using a LAMP-like stack do to improve on things? I kind of thought that dependency problems were inherent to this kind of stack.

E.g. Postmill uses Postgres and PHP. Short of rewriting the PHP part in Rust or Go (which is an extreme length but would produce a mostly-static binary), or using another isolation tool like AppImage or Flatpak, I don't know any way it could be easier to deploy.

I really am asking from curiosity-- I moved to Docker because of frustration with LAMP-likes. (Shaking my fist at Nextcloud)

2