Seems like title cleanup is a good use for an LLM actually. (Not to remove the human option, but as the first smarter pass instead of simple stopword lists.)
An LLM burns through how much power, compared to the tiny Arc server? You could outsource it, of course... But then you have a third party dependency.
Then you have the problem that LLMs continually and constantly change words to be "simpler" and destroy the very sensitive meanings, especially words in tech. Like auto-correcting Fediverse into Universe. Or C into C++.
How does this compare to DataHub? I have been comparing the two and I don't have any pros/cons that really help me pick. What I've noticed:
* It seems like DataHub has an async Kafka ingestion approach while OpenMetadata is API
* It seems that the data model of OpenMetadata can be extended/modified by API calls while DataHub only allows for changes to be made in code (in a forked version, no less) and deployed.
* It seems that DataHub comes with more data types out-of-the-box such as JSON Schema
* It seems like DataHub has an async Kafka ingestion approach while OpenMetadata is API
We do not use Kafka by default. If someone needs kafka they can add it. However for Metadata APIs, we do not feel like Kafka is needed. Lot of projects are getting dependent on Kafka and calling it as real-time. Its unnecessary burden on users who are going to operate in production for 99% of use-cases Kafka is not needed, coming from a Kafka committer :)
2. Yes all of our APIs and Entity definitions are generated using JsonSchema. For us, Json Schema has been awesome, all of our backend / ingestion and UI is generated from JsonSchema and its easy to extend and add new models when needed
I use and strongly prefer a split keyboard for these reasons.
I’m not too far down the rabbit hole but when I got my Ultimate Hacking Keyboard[1] I was finally a happy camper. The improvement over the few membrane split keyboards available was noticeable. Especially once I figured out the layer mappings I like.
It’s sooo much better than the butterfly MacBook keyboard that I basically don’t use my laptop away from my desk any more. I always pack my UHK when I travel. I use a Logitech mouse between the two keyboard halves.
I waited years from the initial find on Crowd Supply to having it. Then several more years before the promised “modules” shipped. By the time the modules arrived I had adapted so completely to the keyboard I haven’t actually found much use for the extra key cluster. I want to like the trackpad module but it doesn’t have two-finger scrolling so it just annoys me when I try to use it.
EDIT: Forgot to mention, this setup resulted in so much less hand, arm, and shoulder pain.
I'm still waiting for a tenkeyless version of the Ultimate Hacking Keyboard. I like the idea of it but chording to get function keys and other keys is not something I can live with.
As a long time vim user I didn’t have as much trouble as I feared for the arrows: just hold my left thumb on mod while using hjkl. (This isn’t quite the native mapping that comes with the keyboard though.)
I mapped mod- and the four right edge keys to home, page up, page down, and end. These were familiar to me from an earlier split keyboard I used. (Goldtouch maybe? Or Kinesis freestyle?)
For function keys, Mod+2 is as easy as fn+F2 on a laptop and not much to remember/learn vs just a raw F2. But I have to admit that some of the shift-alt-Fkey combinations in IntelliJ are even harder to remember and type with an extra mod required.
Probably the thing that really got me was the default keymap requiring mod for escape. So I remapped the corner key to escape, (or backtick with mod) and I have mod-Tab to type tilde. It works, but when going back and forth to the laptop keyboard this is the one that throws me. Also I wish I could teach my Macbook keyboard to have an arrow key layer because I feel like command or caps lock plus hjkl would feel so much better than the tiny inverted T half keys on my laptop.
The UHK is what's termed a "65%" keyboard. A "tenkeyless" keyboard has physical function keys and the arrow key + cursor control cluster; the UHK does not have these things.
Gammu seems like an ideal candidate for a Home Assistant integration or add-on. Could use incoming SMS as triggers and outgoing as notification methods.
A few years ago they wouldn’t let me board a flight from SFO to Australia because my visa said I was born in 1921 when my passport didn’t agree. After a good bit of hassle we just had me go and apply for (and pay for) another visa on the electronic application site and this time I noticed that it’s exactly what happened: 1Password (IIRC) used my Visa expiration year for my visa birthday.
I’m lucky that my status meant it was an automatic approval for the second visa!
It's amazing though that they literally gave a visa to some other random person without any issue at all (i.e. not even cross-checking some values in your passport).
This will really make it a lot easier to move existing apps to a PaaS setup on Mesos. I'm excited that I won't need to do any more port re-mapping, and won't need any more complicated Service Discovery tools. Instead I'll just be able to address my containers using mesos-dns.
This is one of the biggest changes Google added to Kubernetes when moving from Borg, and it eliminates a lot of the complexity in Borg.
