Chris K Wensel

retrofit.shKeep a local PostgreSQL database in sync with the operation log · retrofitSee Retrofit identify which cut a local PostgreSQL database is at from a pg_dump, gate make run on the head of the operation log, migrate the database forward after a git pull, and capture a change ma...

2 Sept

It’s Wednesday, and we’ve got a new demo. retrofit.sh/demos/pull-a...

github.comGitHub - cwensel/intrastate: A Go CLI for explicit, reviewable, deterministic workflow state transitions.A Go CLI for explicit, reviewable, deterministic workflow state transitions. - cwensel/intrastate

If you are looking to improve skill reliability and reduce turns when managing state transitions, check out our new release of intrastate. github.com/cwensel/intr... #llm

@conorpower23.bsky.social

This release improves state lookup against tools that return json, and other minor improvements. In our limited test data (80 pipeline runs, same model, two weeks either side), we're seeing a 33% drop in turns and a 47% drop in context tokens in the resolve step. YMMV.

Chris K Wensel reposted

Vikram Seraph 🪽

@vikramsaraph.com

TIL git switch —orphan

Chris K Wensel reposted

Conor Power

Accepted papers are up and registration is open for CIDR 2027! January 24th-27th in Amsterdam. www.cidrdb.org/cidr2027/pap... www.cidrdb.org/cidr2027/reg...

cidrdb.orgCIDR 2027 - RegistrationThe 17th Conference on Innovative Data Systems Research (CIDR 2027), Registration Information

Chris K Wensel reposted

jake

@yetanotheruseless.com

I actually wonder whether the Autor/Thompson complex Jevons analysis applies to mathematics, and in the coming 5-20y, *more* people will be doing math, rather than less, because it dramatically lowers the barrier to entry, and opens the field, when the degree of required effort / dedication drops?

jake@yetanotheruseless.com

It's a little more subtle than that. I think I mostly agree with Christian Szegedy's point of view on it: docs.google.com/document/d/e...

docs.google.comQuo Vadis, Mathematics?

Chris K Wensel reposted

Zoe Corbyn

@zoecorbyn.bsky.social

Saddened today to hear of the death of Margaret Hamilton. She was amazing and it took some hoops to get an interview with her, but I did in 2019. Nice to see obits in the Guardian, BBC, and the Washington Post drawing from it. Read in full here: www.theguardian.com/technology/2...

Chris K Wensel reposted

Phil Eaton

@eatonphil.bsky.social

Over 10,000 open software jobs now indexed on The Consensus, at nearly 900 companies building software infrastructure. theconsensus.dev/jobs.html

Chris K Wensel reposted

Thorne 🌸 - Play Cozysters! 🎢

@ens0.me

Software engineering is now about being able to make quality decisions around architecture, design, and technology. The actual engineering all along, the code just was in the way.

Chris K Wensel reposted

Dmitriy Ryaboy

@squarecog.bsky.social

Last decade's tools are the wrong paradigm for dealing with the speed and volume of changes happening to this decade's codebases. We need a better changescape: changescape.dev

changescape.devObserving the ChangescapeEngineers increasingly steer how a system evolves rather than write each line. Diff-based PR review is the wrong level for that job. We need a new generation of tools, built to help us quickly unders...

Chris K Wensel reposted

Tom Simonite

@simonite.bsky.social

"AI is steadily reshaping parts of the U.S. economy in ways that aren’t immediately obvious. Zooming in on those changes reveals how the AI investment boom is creating economic dynamism but also instability." via @shiraovide.bsky.social www.washingtonpost.com/technology/2...

washingtonpost.comThe unexpected ways that AI is reshaping the U.S. economyThe technology is bringing new prosperity to certain parts of the economy, but also creating new vulnerabilities.

retrofit.shInitialize a PostgreSQL development or test database from labelled datasets · retrofitCapture baseline seed data from pg_dump into data files under a label, start a postgres container from one rendered SQL file, and layer a fixture for one regression over the baseline by naming both la...

Hey all Continuing from last week's demo on reference data, this week we show off how we handle seed data. retrofit.sh/demos/seed-d... #postgreSQL #sql #llm #dataBS

And if a later table or column rename is recorded in the log, retrofit will create the snapshot with the renames so the datasets do not need to be recaptured with the new names. This works because when we capture the dataset, we also capture where in the change history, the version, it was captured.

Don’t forget to sign up on the site for announcements about early access.

Hey all Continuing from last week's demo on reference data, this week we show off how we handle seed data. retrofit.sh/demos/seed-d... #postgreSQL #sql #llm #dataBS

We see this data as layered. You may have a baseline dataset, but you usually also want to layer on additional fixtures to support different use scenarios or set the foundation for a regression test suite. retrofit lets you compose a final dataset by declaring what is to be included via labels.

The metadata within the datasets also informs the composition. A table in one dataset, for example, may declare that it appends rows to an earlier dataset declaration, or it may outright replace that table. retrofit honors this and composes a snapshot with exactly what was requested.

Hey all Continuing from last week's demo on reference data, this week we show off how we handle seed data. retrofit.sh/demos/seed-d... #postgreSQL #sql #llm #dataBS

Seed data is the data you use to initialize a developer database instance or a CI/CD test suite. We treat reference data as part of the schema, it’s versioned and made tamper evident. Seed data never goes to production, it is a stand-in representative of your production data, warts and all.

retrofit.shManage PostgreSQL reference data with the schema · retrofitCapture a PostgreSQL lookup table's rows from pg_dump into a data file pinned to the operation log, retire the .sql side file that held them, rename the table and its column, and see a hand edit to th...

Hey all Finally, I think we landed on a great design with handling both reference (and seed) data in retrofit. This week we demo managing reference data as part of the schema. retrofit.sh/demos/refere... #postgreSQL #SQL

Additions (or updates) to the reference table are captured in a new file and pinned against the current DDL version. So a migration between the two points in the schema history will result in a script that only applies the changes (adding the new currency to the currency table).

A downgrade will just remove the record(s). If a column or table is renamed that holds the reference data you don’t need to capture it again, the renames will be applied when the snapshot or migration script is rendered.

Reference data is the data that apps only read, and never write, like zip codes and states. This includes data that might change slowly over time, like an order status label, or even currencies. There are a lot of different strategies to manage this data in migration scripts and data files.

Retrofit can capture reference data and pin it to a specific version of your schema. By default, it is included when rendering a snapshot or migration script. If witnessed and sealed, the data cannot ship if changed underneath, retrofit will detect the edit and refuse.

@gergely.pragmaticengineer.com

if we remove toil from work, what's left. pride in our art and craft?

Chris K Wensel reposted

Gergely Orosz

Very much enjoying this “GPU glossary” mini-book by Modal I got from Can Duruk when stopping by at Modal’s NYC office. I don’t know if the physical books can be purchased, but the book has an online version, as I learned; with the same contents: modal.com/gpu-glossary

Chris K Wensel reposted

Brian Sletten

@bsletten.bsky.social

Ahhhhh!!! Sorry, this is for the music geeks. HT @mrbirchtoo.bsky.social #music #morphine #sandman pitchfork.com/story/morphi...

pitchfork.comMorphine Unveil Lost Album CocoonThe album, led by the title track, was completed from recordings shelved soon before the death of frontman Mark Sandman

Chris K Wensel reposted

JD Long

@jdlong.cerebralmastication.com

I’ve literally seen this in the wild. Had exec say they needed “real time data” … I knew they didn’t. Tech team was architecting Kafka bullshit so I asked in a meeting “how many times a day do you need this refreshed?” Answer: “once is fine. Twice if you can”

Samuel Wong@samuelwong.bsky.social

Do You Actually Need Real-Time Data? Why “real-time” is often a business requirement that hasn’t been defined yet seattledataguy.substack.com/p/do-you-act...

seattledataguy.substack.comDo You Actually Need Real-Time Data?Why “real-time” is often a business requirement that hasn’t been defined yet

parquet.apache.orgALP: Adaptive Lossless Floating-Point Encoding in Apache ParquetA technical overview of ALP's design, performance, and adoption across the Apache Parquet ecosystem.

took my this many days to realize jev was named after William Stanley Jevons of Jevons Paradox fame. I guess I'm laxing on my vitamins.

Chris K Wensel reposted

Dmitriy Ryaboy

@squarecog.bsky.social

21 Sept

Kind of embarrassing how many ML people are reacting to Jev with "lol, LLM noobs discovered classifiers / BERT".

Chris K Wensel reposted

Julien Le Dem

@julien.ledem.net

23 Sept

Fresh off the press: ALP: Adaptive Lossless Floating-Point Encoding in Apache Parquet parquet.apache.org/blog/2026/09... Thank you Kosta Tarasov, @andrewlamb1111.bsky.social, Prateek Gaur

Chris K Wensel reposted

Murat (Distributolog)

@muratdemirbas.bsky.social

[new blog post] What Happens When the Model Eats the Stack? Rethinking the Research Agenda for Data Agents to Withstand the Bitter Lesson muratbuffalo.blogspot.com/2026/09/what...

muratbuffalo.blogspot.commuratbuffalo.blogspot.com

Chris K Wensel reposted

acmqueue.bsky.social

@acmqueue.bsky.social

CAFE(S): Your Agent Is Only As Good As Its Context Even the latest models degrade when operating on unclear inputs. We present CAFE(S), a framework that evaluates context across five dimensions: clarity, actionability, fidelity, efficiency, and security queue.acm.org/doi/10.1145/...

retrofit.shAdd foreign keys to a Rails polymorphic association with Retrofit · retrofitSee strong_migrations stop a one-migration refactor of a Rails polymorphic association, Retrofit render the five-migration recipe forward and in reverse from one operation log, and Rails run both dire...

Hey all It’s Wednesday and we have a new retrofit demo to share: retrofit.sh/demos/rails-... This is similar to the Liquibase and Flyway demos. We inject retrofit into the Ruby/Rails developer workflow. Bootstrapping an existing Ruby project into retrofit is quite trivial.

Check out our extended homepage content to learn how retrofit can help you and your team especially with managing risk where other tools leave gaps. Plus, even stronger support for migration reversal in the coming weeks.

As always, we welcome your feedback and requests. The repo is available for early access users.

The win is when you broker all your changes through retrofit, it will render the forward and reverse migration scripts for you that Ruby can run like it always has. One less thing to worry about since Ruby handles so much already but punted the risky stuff back to the developer.

18 Sept

Maybe code was never a craft. That was just something we told ourselves.

18 Sept

I’ve had two things double delivered from Amazon this week. Looks like vibe coding is working out great for me.

retrofit.shReview schema-change branches against a production snapshot without reloading it · retrofitThree branches, one snapshot of a production PostgreSQL database. See make db-up put each branch's schema change on the snapshot from its own dump, make db-down take it back off, a branch that drops a...

17 Sept

Hey all We have a new retrofit demo to share: retrofit.sh/demos/manual... Imagine your team has been using retrofit for the past couple weeks and now every branch to be reviewed has a schema change that will require a migration in production. grin #sql #db #postgreSQL

Finally we merge (rebase!) all the branches, confirm there were no structural conflicts during the merges, and again test the combined up and down migrations. If you or anyone on your team would be interested in getting early access, have them sign up on our website. Please share!

17 Sept

Also, I’m not sure what we can pull off, but making reference data (data in tables your application only reads) first class and versioned with the schema is one of the harder problems we are currently designing and testing. Hope to have a demo of that soon.

In this demo, we show a user manually testing both the up and down migrations against a production shaped snapshot before merging them locally. In one of the branches, retrofit stops a migration because one of the schema changes would result in a dropped column, so it’s skipped.

16 Sept

Current status

Chris K Wensel reposted

Onyx

@advantage87.bsky.social

15 Sept

Replying to @chris.wensel.net

state changes are only as good as the markdown they're written in