Maybe they're starting to realize that the first step to fixing a problem is admitting you have one? Feels safer to say "look we didn't do well and we're going to not do well for a bit, but we'll get back on our feet" because it could encourage a buying of the dip.
I used git bisect in anger for the first time recently and it felt like magic.
Background: We had two functions in the codebase with identical names and nearly identical implementations, the latter having a subtle bug. Somehow both were imported into a particular python script, but the correct one had always overshadowed the incorrect one - that is, until an unrelated effort to apply code formatting standards to the codebase “fixed” the shadowing problem by removing the import of the correct function. Not exactly mind bending - but, we had looked at the change a few times over in GitHub while debugging and couldn’t find a problem with it - not until we knew for sure that was the commit causing the problem did we find the bug.
In the battle the local farmers resisted the Danes as they retreated to the walls of the town where the townsfolk didn’t let them in. They were slaughtered and then the townsfolk surrendered. The mass graves of the farmers have been a major archeological excavation.
We (at my work) are evaluating Iceberg, Hudi, etc. as a next step in scaling up to support larger OLAP workloads beyond what we can currently manage in Redshift and we're leaning towards Iceberg. Does Hudi have a niche that it's especially well suited for?
“Real time” is why we picked it at Notion. I am not a data person so I don’t really know. But I am frequently bummed out when AWS or some other vendor ships a cool integration with Iceberg and/or Delta Lake and there’s nothing for Hudi. It definitely feels like the 3rd place finisher in the popularity contest (Iceberg #1, Delta #2).
I mean, Iceberg won enough that Databricks bought them…
Apply password protection to anything with a dead-simple app that runs entirely client-side. Encryption is handled by RustCrypto and is executed in your browser as WASM.
How does it work? You enter a password, which Argon2 converts to a 256 bit key. That key is then used to encrypt and decrypt via AES256. As a bonus, all plaintext is compressed prior to encryption.
This app was purpose built to delight its creators. We love all feedback and contributions nonetheless!
Very cool and intuitive; if x0 and x1 are in category A, while x2 is in category B, then the concatenation of x0,x1 is more compressible than the concatenation of x0,x2. It makes sense that this would do better than an out of distribution NN.
Simplicity aside, practicality is unclear. It seems you can’t escape the need to perform at least one compression operation, per class, per inference. Eg classifying X into one of 10 categories requires a minimum of 10 string compressions. Probably more if you want better accuracy.
I’ve always wanted a tool that could search ts code using types & logic, e.g. you could search for every instance where an object of a particular type has a value assigned to one of its properties.
.jpg){kind=link}
Why would they willingly start publishing numbers that are worse than 3rd party consensus, out of nowhere?