lbhdc

I am in it for the thrill. Code first ask questions later.

This is me. After I hit ~70 domains, I realized I needed to stop. Now, I have since pared that down considerably, and instead add all of the domains I like to a "wishlist" so I can come back to them (I never do).

Then for my project ideas, I find writing them out to really be what scatches most of the itch for me. I get to think through the problem, think about how I could implement it, maybe even do a little exploring of tech I could use to solve the problem.

For the vast majority of the ideas I don't circle back. For some ideas I will come back a few times and iterate on the plans and designs, but still never build it. And even fewer I actually build. Its all of the fun, without feeling like I can never finish a project. Instead I feel like I can't start them.

I have all of my side projects in a big monorepo managed with bazel. Its got a little bit of everything in it.

All of my UIs are written in typescript, most using Angular but I am switching to a homegrown framework for smaller "finished" projects. I have been looking into adding a GTK desktop app, but haven't determined the language I want to use.

Web servers are predominantly go, but also have c++, java, and rust servers. Everything is grpc with my own frameworks smeared over that.

I put envoy in-front of all of my web servers, and use custom filters for shared functionality across all servers. Things like request telemetry, transcoding grpc to json, auth, and waf are all in this layer. Having it separated in the proxy layer avoids needing to reimplement these functions across the different servers.

For CLIs, virtually everything is in go, but I also have rust, zig, and bash.

ETL stuff is all go. I've looked into using apache beam, but my homegrown framework does what I need, so I haven't made the switch.

Deployments are all to knative. I have a k8s cluster that I run it in at home, and use the managed version on GCP (cloud run). Deploys are all handled by custom bazel rules.

Infra is all managed with terraform (managed by bazel).

I use a few different types of databases. Postgres (or cloudsql) for application relational db. For most things I can use a graph database, and use Firestore for that. Its much cheaper to run Firestore instead of cloudsql for small amounts of data. I am using biguqery for my data warehouse.

Model training is done in jax/tf, and inference is all done with tfserving.

Its a pretty neat idea. Microsoft did that (is doing?) a while back, but I question the practicality. If there is a hardware failure, swapping it out is going to be much more expensive if it requires a ship.

https://news.microsoft.com/source/features/sustainability/pr...

I think this is definitely the future for a lot of developers. Though, like other serverless platforms once you start getting scale this will likely be more expensive than always on variants.

GCP recently announced Cloud Run can do, more or less, what you are proposing. It scales to zero, and you only pay for the request duration.

https://cloud.google.com/run/docs/configuring/services/gpu-b...

When you expand a directory that only contains one directory, it would be nice if it would recursively expand.

Things like this has really got me thinking. If the AI hype all comes to fruition, and you want to ensure good outcomes for yourself, what is the best course of action?

Is it really building an AI company in the hopes that you find something that gets traction? Or would a better plan be building a private military force to take AI from whoever gets it? Would VC want to invest in that as a hedge?

To some extent, yes. I speculate that is why they are sometimes not putting the AI overview first.

They do have ads in search though. I guess it depends on how much more money they make from people spending more time not leaving google vs showing you display ads on other sites.

I think it gets tricky with AI overviews. Some sites are saying they have lost the majority of their traffic (some losing more than 90%). Many of those make money from ads (often by google).

Some sites don't get enough traffic from google to sustain their business where they previously did. Wholesale blocking google crawlers doesn't seem like a risky move for them.

It makes me wonder if that is a trend? Will more sites go to 'google zero'?

My preference is to deploy them to Cloud Run on GCP. Partially because I am already using GCP, but also because its pretty easy to make it always on while staying in the free tier (or very minimal cost).

I figured that was the purpose of the API key, but that puts the onus on me to figure out acquiring that key. Which means I need to also be sold on using the other thing before I can determine if I want to use the first thing.

Having a third site to find a video of the product is even more friction.

It would be helpful if you showed what the output is, or have some queries that don't require an api key.

Currently when you go the the page and search for something, it nags you about an API key for a service that isn't this one (confusing misdirection). There aren't any obvious links until you get the side bar to open.

When you visit this other product page there are some screenshots that aren't very clear what the product is (IDE?) and the description above the fold just says something about supercharging ai-agents or something (effectively saying nothing).

That is a lot of friction just to see if this is something that solves my problem.

While that sounds fun, I think this misses that egress (typically) isn't free. Returning 403 and no content is cheaper. If a crawler is generating a lot of traffic dynamically generating content will make that crawl even more expensive for the operator.

As far as I am aware it is not a derivative of another database.

https://dbdb.io/db/clickhouse

Its a db company that offers an open source database and cloud managed services.

The database is OLAP where Postgres is an OLTP database. Essentially it very fast at complex queries, and is targeted at analytics workloads.

Thanks! I am gonna give this a try.

There is a nice library that abstracts a few core things like secret management for the major cloud platforms and some self hosted or local options.

https://gocloud.dev/howto/secrets/

I am in a pretty similar situation as you, and have really been feeling the benefits of going all in on bazel.

> We have made this tool layer which means if I run `go` or `kubectl` while in our repo, it's built and provided by Bazel itself. This means that all of us are always on the same version of tools, and we never have to maintain local installations.

Currently I have to run `bazel run <tool>`. Your solution sounds way better. How does yours work?

What is the value of disappearing posts in that context? I've never used snapchat, but I have used reddit, so I am not sure what the appeal of snapchat is.

That seems pretty far off from my experience. They offer auto pay, so I have paid the balance on my cc every month since I have had it without ever accruing interest. I haven't be denied other cards from them, balance increases, or anything like that.

There was an interesting video by the Religion for Breakfast channel that talked about the process and where some of the rules came from.

https://www.youtube.com/watch?v=DNwgh787umM

> What's in it for Oracle?

Customer acquisition, like all companies with a free tier. It lets people experiment with their products and see if it meets their needs. Maybe those experiments grow up to be real products and continue running where they are. Maybe that user becomes an advocate for that product to their employer.

I run a number of go servers with similar resource requirements on cloud run for free. I wanted them to be always on, so I set up an uptime check that keeps those instances alive around the clock without leaving the free tier.

Good catch. Its weird that it compiles without error as a consteval func.

You can use `std::bit_cast` to do that in constexpr contexts.

    constexpr auto f(uint8_t *x) {
      return std::bit_cast<char *>(x);
    }

https://godbolt.org/z/K3f9b9GGs

This compiles in clang 21.1.2 with c++26.

    #include <cstdint>
    #include <print>

    constexpr uint8_t f(char ch) {
     return static_cast<uint8_t>(ch);
    }

    int main() {
     constexpr uint8_t r = f('a');
     std::print("{}", r);
    }

Such an underrated movie. Great watch for anyone interested in classic scifi.

For big companies like this, I like to go look at the activity on sdks and related projects. Typically activity will flat line before a sun setting announcement is made. Its hard to say how much lead time that buys you, but I have found it a useful barometer.

It seems like the firebase repos have had a fair amount of activity.

https://github.com/orgs/firebase/repositories

Is there an alternate link? This is just an ad for their premium sub.

My flow is more or less this:

git merge -> ci -> oci artifact -> cd -> cloud.

Every deployable is packaged as an image, and can be deployed to serverless runtimes available on many clouds, VMs, and k8s (I assume other orchestrators too, but haven't tried).

My goal is to commoditize my cloud provider, while minimizing my costs. Everything is configured through terraform, so standing up an equivalent environment on a new cloud is pretty trivial.

I've tried to be very mindful about what I depend on from the provider (eg using provider specific sdks). I have had mixed results at sticking to this. I would like to improve this to the point where I could automatically fail over to other providers.