
Paul Dix
@pauldix
Followers
9K
Following
1K
Media
135
Statuses
6K
CTO of @InfluxDB (YC W13), founder of NYC Machine Learning, series editor for Addison Wesley's Data & Analytics, author of Service Oriented Design with Ruby.
New York City, NY
Joined February 2008
Prateek Gaur and co at @Snowflake reproduced the (great) results for the ALP encoding algorithm from @cwi_da / @afroozeh3 / @peterabcz. ALP achieves ZSTD levels of compression and much faster decode. We are discussing adding it to @ApacheParquet: https://t.co/gxwF5QqtNO
0
10
77
Our announcement about @InfluxDB 3 Core and Enterprise availability in AWS Timestream:
influxdata.com
InfluxDB 3 Core (open source) and InfluxDB 3 Enterprise are now available as fully-managed services on Amazon Timestream for InfluxDB, giving developers a direct path to deploy and scale real-time...
0
0
2
AWS announcement about Timestream for InfluxDB 3 availability:
aws.amazon.com
Discover more about what's new at AWS with Amazon Timestream now supports InfluxDB 3
1
0
1
Big stuff today: Amazon Timestream for InfluxDB now supports InfluxDB 3 Core and Enterprise! This continues our partnership with AWS to bring @InfluxDB to as many developers as possible. This is just the beginning, we have big updates coming over the next six months. Linksβ¦
1
0
5
Great read on SOTA coding with agents. The commentary on Codex being better matches my experience. I'm still using different terminal windows and branches for multiple agent work. I'm working in Rust so I'm not sure the "everyone in the same branch" would work for me.
π’ Time for an update on my workflow. This one's a 23 min read, so buckle up. 100% organic and hand-written, like an animal.
1
0
1
The pharmaceutical fallacy lies with the idea that everyone claims price controls will eliminate profitability and decrease the profit motive that fuels the growth of medicines. This is only partially true because a profit motive is required, but how much? This contributes to the
2
2
22
Update on this Codex vs. Claude competition: Codex High was the clear winner. But Sonnet 4.5 did find one thing that Codex didn't uncover. Both of them got hopelessly lost in the beginning until I pointed at a couple of commits that I thought could be the source of the problem.
Investigating a bug, I have GPT-5-Codex High in one terminal and Claude Sonnet 4.5 in another competing to see which can solve it first. Maybe one, maybe both, maybe neither and I have to actually figure it out for myself. First step, create a reproducer test...
0
0
1
The @CMUDB seminar series talks today looks like a great one. @_willmanning presenting on @vortexdotdev, the columnar file format the folks @SpiralDB contributed to the Linux Foundation. Should be interesting viewing for all file format enthusiasts:
db.cs.cmu.edu
Apache Parquet revolutionized columnar storage after its initial release in 2013, but... Read More +
0
6
34
Investigating a bug, I have GPT-5-Codex High in one terminal and Claude Sonnet 4.5 in another competing to see which can solve it first. Maybe one, maybe both, maybe neither and I have to actually figure it out for myself. First step, create a reproducer test...
1
0
1
More exciting work in file format land. And a little bit of backstory for those like me that live for tech nerd drama.
Our SIGMOD paper with @XinyuZeng218 + @huanchenzhang + @wesmckinn + @pateljm on creating a next generation open-source data file format is out. F3 is a future-proof file format avoids the mistakes of Parquet. π Paper: https://t.co/fnFwN9gxbZ π Code: https://t.co/R5nhlVvsea
1
0
6
Congrats, Andrew! Looks like Parquet is going to be getting FSST support? π
I am proud to announce I am now a committer on the @ApacheParquet project. Realistically this likely means more reviews / helping clarify the parquet specs, but I also hope to help more actively evolve the format, especially around new encodings. https://t.co/lnR71Po1yA
1
0
7
So far, GPT-5-Codex is cranking. Going to keep this thing working overtime to see if it gets somewhere.
I switched over to using GPT5-Codex & CLI as my daily dev driver this week. The UX isn't quite as polished as Claude Code, but so far I'm very impressed. The results have been quite good. Going to toss something at it that Claude has been struggling with to see how it goes.
0
0
2
Note that with Claude Code I only use Opus. I don't bother spending my time with anything but the SOTA frontier models. So with Codex, I'll only use GTP5-Codex until they upgrade. I put it on medium thinking because that sounds like it's the sweet spot for getting best results.
1
0
2
I switched over to using GPT5-Codex & CLI as my daily dev driver this week. The UX isn't quite as polished as Claude Code, but so far I'm very impressed. The results have been quite good. Going to toss something at it that Claude has been struggling with to see how it goes.
2
0
3
I forgot to post about this nearly 2 hour podcast I did for Developer Voices that went up last month. I go into pretty deep detail about @InfluxDB v1 and TSM, rewriting in #rustlang, building with @ApacheDataFusio, and the perils of a big rewrite:
0
0
6
Huge, congrats! Can't wait to see you and your team do even more cool stuff. And to work with @vortexdotdev!
Proud of what we've built, but even more excited for what we're going to do next π @SpiralDB @vortexdotdev @generalcatalyst @AmplifyPartners
0
1
4
Overheard in a Slack channel thread about Claude Code being down: "I guess I'm typing in code like a peasant." LOL, indeed.
1
0
6
A piece I wrote about rebuilding InfluxDB in #rustlang and #ApacheArrow and #ApacheDataFusion is up on InfoQ!
Engineering a Time Series Database Using Open Source: Rebuilding InfluxDB 3 in Apache Arrow and Rust https://t.co/unSvRB0sS9 authored by @pauldix, reviewed by @olimpiupop
0
0
6
Even the Postgres ecosystem is getting in on the awesomeness that is DataFusion (@ApacheDataFusio) https://t.co/Pite8Lx4qW Exciting to see adoption in established databases!
0
0
1
A deep dive on wal3 - Chroma's open-source write-ahead log built on object storage. Featuring: - a 30-year-old lock-free algorithm - Amazon S3's newest conditional writes feature - a novel checksumming technique called setsum
8
27
194