rupurt

I'm on a similar journey. I took flight 6 weeks ago and built a turn based board engine for human/agent delivery teams called Keel https://www.spoke.sh/keel. The grand vision is to apply the board engine as a control mechanism for work to be done and verified in deployed robot fleets.

I built an agent observability & orchestration toolkit over the holidays https://github.com/run-vibes/vibes. It uses Iggy as the message streaming layer and currently supports PTY streaming for Claude Code to the browser, Cloudflare tunnel, automatic Claude Code hook instrumentation into Iggy and I'm using that to build a cross harness continual learning system that I'm hoping will act as error correction for the model environment. It's using Iggy server v0.6 (io_uring on Linux) and compiles to native platform binaries for Linux & Mac (Windows support is on the list...)

As other folks have described programming on the mainframe is about learning COBOL + JCL. I would also add that you need to understand how to read/write binary files according to a defined schema (Copybook) and encoding (EBCDIC, COMP, COMP-3, COMP-5). It's also helpful to deeply understand the intricacies of Db2 and different methods of connection (ODBC, JDBC, Db2 CLI).

I've even gone so far as to write an ODBC extension for DuckDB so that we can scan and process queries faster and with less MIPS usage. https://github.com/rupurt/odbc-scanner-duckdb-extension

https://www.mechanical-orchard.com

FWIW I came to a similar conclusion that a lot of the power in Kafka comes from the API and that eventually much of the complexity of managing the cluster will eventually be abstracted away with multiple implementations. I also felt that if I could implement Kafka persistent over the S3 keyspace then I could start with persistence direct to S3 like you've done with warpstream and then layer on a faster hot disk and in memory tiering mechanism to eventually lower end to end latencies.

I love where you're going with this so hit me up on twitter if you ever want to chat more in-depth https://twitter.com/rupurt.

It's now on the list of improvements!

It supports fetching rowsets in batches to minimize network overhead and defaults to the default DuckDB vector size of 2048.

I've tested it against the IBM DB2 & Postgres ODBC drivers and will continue to test and add support for all major databases. If you've got one you'd like to see let me know in the comments.

I've got plenty of improvements in the pipeline including: - Write to ODBC database - Predicate push down - Automatic catalog discovery - Performance benchmarks - Configurable batch size - Relative & absolute cursors - OS X builds using rosetta for databases without native aarch64 drivers (IBM/Oracle)

  <div
   phx-hook="MyChart"
   data-days="<%= Jason.encode!(days(@balances)) %>"
   data-amounts="<%= Jason.encode!(amounts(@balances)) %>"
  >
   <div phx-update="ignore">
   </div>
  </div>

There are a number of forks/patches out there that provide hasOne semantics and saving for multiple records in the same commit (ours are the mhelabs ones)

- https://github.com/mhelabs/ember-data/tree/has_one

- https://github.com/mhelabs/ember-data/tree/parent-child-comm...

- https://github.com/ghempton/data/tree/relational-adapter

Based in sunny Perth, Australia :)

We have high-res PhotoMaps of the major Australian cities and allow you to see change over time by flicking through the dates of places we have captured.

http://www.nearmap.com/?ll=-31.959417,115.859435&z=17...