Article URL: https://support.apple.com/en-us/149555 Comments URL: https://news.ycombinator.com/item?id=50271685 Points: 17 # Comments: 3
This is an AI-generated summary. ShortSingh links to the original source for the complete article.
Factry Historian stores its time-series data in InfluxDB v1.x and v2.x, with configuration and metadata kept in PostgreSQL, so the real choice is between a finished process historian and a self-assembled one. Factry sells a complete product with collectors, asset models, event processing and a Grafana plugin for a flat fee per site, while InfluxDB is a database that requires users to add collection, modeling and dashboards, such as through Telegraf. Pricing differs: Factry does not publish historian prices and has no tag limits, whereas InfluxDB 3 Core is free and open source, with Enterprise and Cloud options priced separately. The comparison also notes that InfluxDB 3 is not listed among Factry's supported back ends, though Core accepts the 1.x and 2.x write APIs. The article suggests Factry for a ready-made solution and InfluxDB for teams able to build and maintain their own data stack.
Ankur has built Potjar, a free Android app for splitting bills on trips, among flatmates, and at dinners. The app supports splitting costs equally, by exact amounts, by shares, or by percentages, and computes a settle-up plan that minimizes the number of payments for groups of up to 12 people. It requires no account, shows no ads, includes no analytics, and has no internet permission, so data stays on the user's phone unless exported or shared, and it never moves money. Potjar is currently in Google Play closed testing rather than on the public Play Store, and its creator is seeking honest feedback from testers.
A developer's SQL query ran without errors, yet one customer's refund total came out at $60 when only $30 had been refunded. The cause was a JOIN that matched one customer against multiple rows, so the results multiplied rather than staying one row per customer. The article reproduces the issue with a small sample dataset of customers, orders, and refunds for an online store. It explains that a JOIN's result grain can change when it links to the many side of a one-to-many relationship. The author notes that the mistake comes from forgetting that multiplication occurred, not from a syntax problem.
Over the past year, Instagram, Facebook, Threads, YouTube, and TikTok have each added features letting users type the content they want in their own words. The author argues this shifts recommendation away from purely inferred behavior toward explicit requests that posts must match. A declared topic is treated as a strong signal, so posts whose subject can only be described in jargon may be excluded from matching. The author notes platforms have not disclosed how much weight declared preferences carry or how many users actually use them. The piece suggests creators name their topics the way audiences would, in a single clear topic per post.
Discussion (0)
Log in to join the discussion and vote.
Log in