sevazhidkov

It’s not a traditional blog, but Oxide’s RFDs cover exactly what you asked — implementation details and trade-offs: https://rfd.shared.oxide.computer/

To be fair, the most interesting bugs that can be caught with tests always feel like “I would’ve never guessed that part of system actually depends on this one”.

My friend and I had a similar idea a few years ago, so we’ve built a prototype of a tool that converts personal data exports to a single SQLite database: https://github.com/bionic/bionic (repo includes “popular Spotify songs when I’m in transit according to Google Maps” query). Unfortunately, we haven’t found ourselves actually using the aggregated data: we’ve looked on it a few times, but it didn’t end up solving some real pain. It was fun to build though!

Big fan of this series of lectures: https://youtu.be/h-94UhJLeck?si=0bEeB9VjmwRXIWTB.

I’ve recently listened to a talk from Jane Street about their review system: they use this method exactly with a helper tool that keeps track of what you’ve already reviewed in the PR, minus any changes that were committed after review. PR author also “reviews” the branch, so the code review comments and changes would show up on self-re-review (because they were not reviewed).

https://youtu.be/MUqvXHEjmus Transcript: https://www.janestreet.com/tech-talks/janestreet-code-review...

It's very hard to understand what Foundry actually is from the provided landing page. Would you be able to describe it? Is it like Retool, but with more data sources and blocks? Can you actually drag-n-drop new applications in it or it's customized by Palantir only?

If you are interested in tinkering with personal data exports using SQL, my friend and I made Bionic: https://github.com/bionic/bionic.

The README includes an example of calculating songs you often listen to while walking/driving/using public transit (by combining Google Maps and Spotify data).

The problem with these monetary estimates is that they involve current valuations, which are probabilistically based on future outcomes, not enterprise value.

PG once formalized YC value prop as average outcome multiplier [1]. Whatever the current valuation, would YC improve your expected outcome over lifetime of the company on 7.5%? If yes, then you should do it.

[1] http://www.paulgraham.com/equity.html?viewfullsite=1

Given that enough people use this heuristic, there will be companies focusing on earning karma on HN, writing comments and voting for products they are getting paid for.

I believe Hacker News’s backend is written on a Lisp (or, more precisely, Arc). In this context, JavaScript, Python and C are being debated for a role of client-side language handling DOM events.

Also, C and Lisp are different breeds. It’s wrong to put them in one bucket of “greybeard” languages. Some Lisps gained traction in the front end space (see ClojureScript).

Does something like Automattic's P2 platform exist as a standalone product (preferably, open source)? If not, I'd be interested to create something like this.

Presumably, things like Workplace and Basecamp support it, but they seem too heavy.

My friend Alexey (@shekhirin here) and I often make small projects analyzing our personal data. For example, we counted popularity of words over time in our Telegram group chats and analyzed browsing history to determine the most productive and unproductive time for each of us.

Most of the time, loading and formatting the data takes more time than actually finding insights. Consumer services provide a machine-readable data export to comply with GDPR. However, you still have to load into memory, extract relationships and otherwise prepare data before starting to look for insights.

So, we decided to build a tool unifying export formats by various services into a single SQL schema. The tool currently supports some parts of Google, Apple Health, Spotify, Telegram, RescueTime, Instagram, Twitter and Netflix exports.

Bionic is a CLI built with Go. Under the hood, it consists of modular “exporters”, each responsible to load data from a certain service export (for example, Google Takeout zip archive) to an SQLite database. You can share one SQLite database for all exports, making it a central hub for your personal data.

One cool outcome of a single database is that you can easily join multiple data exports and produce a report like “Songs I listen [Spotify] in different locations [Google Maps]”.

Would love to hear your feedback! Also, if the project interests you, consider contributing: for example, by developing new exporters or extending the existing ones (currently, Google exporter covers data only for 3 out of 20+ Google products). Alexey and I are available to help.

Thank you! At the moment, no, but we are planning to create inline mode soon