ngrilly

You're right, but what pron is saying is that it doesn't matter much from a security standpoint because they can't be exploited.

That's scary! And wrong...

Mostly agree, but then you have to take care of replication, failover, and backup. By using a replicated block storage, you can mostly externalize that problem.

Yes, but in practice they are extremely hard to exploit. It has been discussed extensively here on HN and in other forums.

That's because Zig offers spatial memory safety (e.g. buffer overflows and index out of bounds), but no temporal memory safety (e.g. use-after-free). I suppose the "and" before "memory safety" is a typo.

Very cool.

Does it support interior pointers?

I'm in the same situation. But I was shocked discovering it goes both ways: many new Gemini functionalities are only accessible using a consumer account instead of a Workspace account. Also, Gemini is now the only major AI assistant with no support for MCP connectors. Instead of adding this to the core product, like ChatGPT and Claude did, somebody at Google decided that it was smarter to add this fundamental feature to a new product instead: for enterprises this is Gemini Enterprise (which is a product completely different from the Gemini app); for consumers this the new Gemini Spark agent (meaning that you can use MCP within Spark but not within a "non-agentic" chat)... It's clear to me this a symptom of Google shipping their org chart, which is a disaster from a product perspective.

Also a big proponent of Google and Gemini, but their stubbornness in artificially splitting their consumer and enterprise products is extremely annoying. It's pretty weird that I have access to more powerful tools when using my personal Google account compared to my corporate Google Workspace account.

uv is solving these problems for the kind of things Scarf is developing.

Same. Almost everything works out of the box, with great defaults.

But I don't think you can implement RAII in Zig?

Yes, I'm so happy it has been accepted. It will particularly useful to pass named parameters to functions.

> The only thing I'm jealous of with regards to python is how seamlessly you can do JSON schema enforcement on HTTP endpoints.

Yes, it is much easier in Python because type annotations are reflected at runtime.

Python type annotations are available at runtime, which can be used for serialization and validation (for example in FastAPI).