tjake

YouTube interview with creators: https://www.youtube.com/watch?v=gJXZZJZc5gI

You can demo this here: https://jvectordemo.com:8443/

GH Project: https://github.com/jbellis/jvector

it's 35M 1024 vectors Plus the text

The default for `convert.py` is F32. This is just SIMD CPU comparison.

Jlama uses the vector api in java20 but also better thread scheduling with work stealing and zero allocation.

GH: https://github.com/tjake/Jlama

Missed AstraDB (Cassandra based). It ticks all the green boxes (HIPPA, SOC II, Cloud, in DB etc..)

https://docs.datastax.com/en/astra-serverless/docs/vector-se...

This is close to my heart, I love these integrations!

sounds cool, except before you even got close close to the black hole you would be ripped apart by the tidal forces.

The CLA protects users of the software from contributors coming back later and saying: "Hey I own that code and now you need to pay me!" Apache projects in general rely on CLA documents.

It's totally in the right of the contributor to refuse to sign this document. But don't expect your contributions to be applied, or work on any Apache licensed projects :)

I'm not sure though what you mean by "tied together".

This is the issue that even zookeeper has.

Agreed. They discuss this in the chubby paper as well.

There is a grace period where the master is in jeopardy of failing over vs failing over straight away. I left this out for simplicity of the state machine but I pointed to the chubby paper as a true reference.

They also mention the fact that client and master clocks must advance no faster than a known constant factor.

bq. I also see Cassandra timeouts while writing with consistency ANY, yet are still somehow getting timeouts with this operation. That really screams to me that the cluster is truly messed up.

This was a issue with atomic batches and CL.ANY we've fixed since the test.

https://issues.apache.org/jira/browse/CASSANDRA-5967

Yeah I thought the A.I. categories would get some feedback like this. It's a huge field.

Tell him about it :)

Oh sorry. I fixed it. Sorry Geoffrey!

It's free for production use now...

You can either allow traffic between one or more subnets, or put gossip traffic on a private network.

For linux users this is the equivalent: https://github.com/duncs/clusterssh

The tool used to ssh in parallel is http://code.google.com/p/csshx/

The site lets you search twitter lists or the people you follow on twitter, rather than all of twitter.

see the homepage http://sparse.ly

thats C not C++. geesh :P

The only way to access your personal twitter stream is via authenticated access, like any other twitter client.

twitter api limits us to your history, we could resort to scraping perhaps?

yes we index your tweets so you can search them. if you never come back to the site we stop indexing them.

You can revoke access to the app from here http://twitter.com/account/connections

We don't yet support full delete of all history, but it's on our list.

Also, we are adding support for an api so you can interact with the site/export your history programatically.

this does search history from the point you create your account :) thx for the feedback.

Yeah, I know what you mean. I'm in about the same situation. But I save the beer for the weekends and code :)

watch it happen: http://twitter.com/linux_log

... That better have High-Speed Wifi!

yeah, no tunes, and i just couldn't get used to the sweaty hands from the gloves. i forgot about the mask. those dark red lines in your face, i much prefer my aeron chair and office now :)

Wrote custom test-set software in a clean-room. Bunny suit, mask and rubber gloves for 8 hours a day. It was a windows box connected to a x-ray diffraction machine so I had to be in the room.