gerwim

Hehe, looks nice. My take on it is ForgeBoard (name will be changed soon though): https://forgeboard.gerwim.com/

How do you handle the context? As in: you have a forum like system as you say, how are agents finding the correct "posts"? How do you handle drift or specs that change over time?

Launched by CERT? I can't find anywhere (CERT website, github) these are related. Do you have sources?

That would work for a single instance, but when running multiple (e.g., you are horizontally scaled) it would not be ideal.

Is it possible to mimic "subtrace.dev"? There's the 'SUBTRACE_ENDPOINT' environment variable which can be used to set the target endpoint, but is the server side open source too? And does the license grant permission for self hosting the full stack?

Looks great! Reading through the docs it seems the subtrace process sends all data to your server. I'm reluctant to do that on a production environment, where API keys and personal data are being handled.

Is there any way to run it completely self hosted? If not, are there plans? And how will you monitize self hosted options (if it's possible)?

Linux gaming is great. Running Bazzite myself, but for proper support you should run full AMD (CPU and GPU). Nvidia drivers seem to be a pain in the arse.

Core only has the "full backup". Incremental and other types are available to enterprise. I run the Core edition (with full backups) for my personal projects.

Agreed, Immich is awesome and great. It can also do searches like "woman smiling", because all images are scanned by an ML agent as well.

Cool!

There seem to be a lot of projects in this space -- including my own tool (https://github.com/gerwim/tunnlr). Still working on it though! ;-)

> Wyze blamed "a third-party caching client library that was recently integrated into our system" for the trouble.

Yes, of course. Blame a third party library which was probably created by an open source maintainer instead of testing your own systems.

Maybe there’s proper support soon in AI landscape [0].

[0]: https://news.ycombinator.com/item?id=39344815

Another fun fact about .dev which I recently found out while working on a side project:

You NEED to use https when visiting any .dev domain. Google has put it on the HSTS preload list.

It took me a while to find out why my browser kept redirecting me to https when I wanted to use http (local development). Curl worked fine…

Maybe not for Google, but it's not part of the OAuth spec. It's perfectly valid for refresh tokens to expire.

Are you sure? According to Apple [0] it requires your Apple ID password. Also, I can't reproduce this on my iOS device: it's locked by a timer which increases exponentially.

[0]: https://support.apple.com/en-gb/102677

Finally. For current iOS versions, there is a workaround: use the “screentime” feature to disallow pincode changes.

In screentime you can set a different code, so when anyone else can access your phone, they can’t change the code and lock you out of your phone.

You need version 2023.10 (or higher) for both extension and server for it to work.

What is this post ranting about? Pushing your keys to big tech? There is no difference if you use Chrome’s password manager for passwords or passkeys.

You can save passkeys on your own network if you use Bitwarden or if you want, write your own solution.

For 99% of the internet users, passkeys are much better than passwords.

Depending on your platform, but if you read the readme, there’s a pre packaged release with Python embedded.

I use Phind[0]. It's great addition to the regular search engine.

[0]: https://phind.com

If you create a new profile for the different user, you can start a second session:

firefox -new-instance -P "Another Profile"

I didn’t know they needed developers. I’m an avid Plex user but C# developer by trade. Cloned the server and web projects and worked on my first issue!

Which is funny, because I own an iPad from 2018. Which has the A10 chip (which was used in the iPhone 7). Guess what, the iPhone does not get iOS 17 but the iPad does get iPadOS 17.

I did some geekbench tests.

4 GB, 3 CPU AMD [1]

8 GB, 4 CPU AMD [2]

8 GB, 2 CPU (dedicated) AMD [3]

So even though the dedicated CPU option has stable performance, the shared CPU is almost 4x as fast. Sure, the shared results would fluctuate but given they are similar priced. Or maybe I was just lucky and got on an empty node.

[1]: https://browser.geekbench.com/v6/cpu/2360706

[2]: https://browser.geekbench.com/v6/cpu/2360698

[3]: https://browser.geekbench.com/v6/cpu/2360730

Great video [1] on how some clever tricks are used to stay within memory constraints.

[1]: https://www.youtube.com/watch?v=sw0VfmXKq54

Passkeys are not tied to hardware. You can save them in iCloud, Bitwarden and lots of other tools. You can also share them to another device [1]

[1] https://developer.apple.com/passkeys/

Looks great! I see your tool uses Hetzner block volumes for storage. How do you manage your data backups (as Hetzner does not offer anything for their block storage volumes as far as I know).

My current setup is also on Hetzner. 1 master and 3 worker nodes. Managed by Rancher and I use Longhorn (also from Rancher Labs) for storage. It runs stable for two years now. But looking for a simpler solution, so I looked into k3s and k0s last week -- will also look into your tool ;-).

Wasabi's egress is not fully free as you make it sound. Their egress works like this: if you store 1 TB of data, you have 1 TB of free egress, as seen in their FAQ [0]. This means it's unfit for certain cases and another competitor could be a better option.

[0]: https://wasabi.com/paygo-pricing-faq/

While you are technically correct, kubernetes removed the Dockershim. Docker is often used as a synonym for containers. However, most of the things should still work [1].

[1]: https://kubernetes.io/blog/2020/12/02/dont-panic-kubernetes-...

Must be HN hugged to death. API responses are returning status code 500.

Looks great! I currently use Auth0 for my projects. One specific feature in Auth0 prevents me from switching to any other IAM: rules (or their new version: actions).

It’s basically running some JavaScript upon logging in (or signing up, or when a machine to machine token is created). It allows me to extend the created JWT with custom properties and perform backend logic (e.g. perform an API call upon user registration).

Is this something Logto will eventually support?

Well there is Dolt: https://www.dolthub.com/