Varun Thacker

Search and Discovery at @SlackHQ. Apache Lucene/Solr committer. ex @Lucidworks @Unbxd

[matt rothenberg @mattrothenberg]

just picked up this bad boy. can't wait to write some software with it

[Andrew Lamb @andrewlamb1111]

You can use @ApacheParquet for Vector Search with embedded indexes:

> We don’t change the file format; we just tune it.

@MOVNTDQ explains how in https://t.co/39IMeYYO7A

[Varun Thacker @varunthacker]

@Qagggy I had a fun time at https://t.co/cay1xAjcBH . they have a ton of beer to chose from and drink there. "Dr beer" can help you choose or you can do a tasting session.

[Varun Thacker @varunthacker]

Tried out JDK21's new Structured Concurrency APIs https://t.co/maEU9RKUMs to demonstrate a fan out scenario. I'd love feedback on how to incorporate timeouts correctly.

[Varun Thacker @varunthacker]

@FlaxSearch https://t.co/YLM1qMVeX2 sheds more light

[Debasish (দেবাশিস্) Ghosh 🇮🇳 @debasishg]

TIL: AstraDB / Cassandra started with HNSW algorithm for vector indexing, which is fast, accurate and easy to implement. But it needs a lot of memory. Hence they switched on to DiskANN that uses limited memory and a more compact representation of vectors, support incremental indexing (as opposed to static indexes) and offers lower latency on SSDs.

AFAIK many other vector search engines like Qdrant uses HNSW for indexing.

Here's the implementation for Cassandra / AstraDB - https://t.co/iq498UyJ6m

Here are the links for the papers that discuss both the algorithms:

1. Efficient and robust approximate nearest neighbor search using Hierarchical Navigable Small World graphs - https://t.co/UBoaPboRi0

2. FreshDiskANN: A Fast and Accurate Graph-Based ANN Index for Streaming Similarity Search - https://t.co/i907YjKTU5

[Varun Thacker @varunthacker]

@FlaxSearch From the blog if I click on the linked CASSANDRA Jira(CASSANDRA-18557) it takes me to CEP-30 it talks about "The approach to implementing approximate nearest neighbor (ANN) vector search in Cassandra involves the following steps:"

Maybe that was the earlier approach?

[Varun Thacker @varunthacker]

@AlexKlibisz 👌looks good! Appreciate the nice detailed blog write up! Have a nice weekend

[Varun Thacker @varunthacker]

@AlexKlibisz Sorry I am running into some SBT issues ( haven't tried a scala setup locally in a while ) but hopefully this clarifies it https://t.co/jlSumflE3u

[Varun Thacker @varunthacker]

@AlexKlibisz Sorry I don't think I was able to convey what I was trying to say in a tweet form. Lemme send you a quick test case in a PR

[Jean-Philippe Bempel @jpbempel.bsky.social @jpbempel]

My slides from my talk Mastering GC at #springio23 are here: https://t.co/VNi9G6hfwp

[Alex P @ifesdjeen]

Database Internals @therealdatabass is now also available in audio format, from https://t.co/9mzmCksRfN as well as Audible. Retweet to help spread the news and enter the raffle to get it for free (via redeemable code).

https://t.co/LKEpRSC87i

[Varun Thacker @varunthacker]

@FlySWISS is flight LX 155 from BOM to ZRH cancelled for today, i.e., MON 26-Dec? I didn’t receive any notification of the flight cancellation / rescheduling. However, I now see the flight operates next on WED 28-Dec. Please help clarify.

[Mike Kowalski @mikemybytes]

I've spent a couple of days with the HotSpot Virtual Machine Garbage Collection Tuning Guide (writing a new article). My 3 conclusions:
- JVM is a piece of art 🤯
- Usually, the answer is: "it depends" 😵‍💫
- Many devs haven't heard about the "Ergonomics" 👂https://t.co/gwMIZ6imnL

[Norman Maurer @normanmaurer]

Async profiler keeps improving and improving … @AndreiPangin is on fire 🔥🔥🔥🔥🔥

[Ben Thompson @benthompson]

This thread is insane.

https://t.co/f2DyYYZkfk

[Varun Thacker @varunthacker]

Water is the messenger that’s delivering the bad news about climate change to your city, to your front door

[ProPublica @propublica]

"My fear is that groundwater will, as usual, be left out of the discussion — groundwater is mostly unprotected, and it’s going to be a real shit show."

https://t.co/5ZEBrdqUXB

[Mark Reinhold @mreinhold]

Here comes Loom!

[Qagggy! @Qagggy]

Dear City: Don’t ask me where I want bike lanes. The answer is everywhere. You don’t ask me where I want water, electricity, and garbage service.

[Varun Thacker @varunthacker]

It's 2022 and I still don't get how google docs hasn't added emoji support ¯\_(ツ)_/¯