morgante

Agents can work in production, but usually only when they are closer to "workflows" that are very targeted to a specific use case.

He mentions someone taking it had joined only a few days earlier.

Awesome to see another open source player in the space, especially after Sourcegraph went closed source.

It looks like you're working on this full-time (and it's a lot of work to build great code search, as I know from working on my own product).

What are your plans for monetizing / building a sustainable business without inevitably going closed source like Sourcegraph?

This has to be a parody.

"Source-available but not strictly OSS" is, in most cases, not really giving anything of value to the community. You are not free to use it in building your own solutions. At best, it allows easier collaboration with customers.

Open core absolutely is open source. There's a clear and valuable open source core that others can build on to build their own products.

This isn't just hypothetical. As far as I know, no source-available license allows you to actually use it to build your own product. While VS Code has many other products built on the core (including products that lawyers have reviewed closely for compliance).

The clearest way to understand it is that VS Code is open core software.

There's actually a lot in that core, especially when you compare it to other open core products (ex. Gitlab). Enough that you can fork it and have a viable MVP for your own IDE (as Cursor, Gitpod, and many others have done)

But you shouldn't ever be under the impression that you fork will be identical to VS Code.

Frankly I think developers have benefited a lot from this. There's actually a strong foundation for others to innovate on their own IDEs without sinking years into basic R&D, and it's much better than the alternative era when ~all the user-friendly IDEs were proprietary.

Nice (simple) introduction to the tree sitter APIs.

If you're looking for a higher level interface, GritQL[0] is built on top of tree-sitter and could handle the same refactor with this query:

  language python

  `def $_($_): $_` as $func where $func <: contains `database` => `db`

[0] https://github.com/getgrit/gritql

This article is likely fake, since it's not on vjeux's actual blog[0] and there's no mention on Twitter.[1]

The style is also all off.

[0] https://blog.vjeux.com/ [1] https://x.com/Vjeux

To be clear, you don't need to make it more complex / failure-prone. I didn't say failover needs to be automated.

Kubernetes or complex cloud services are not required to have some basic deployment automation.

You can do it with a simple bash script if you need to. It's just pretty surprising to see the reaction to a hardware failure being to wait around for it to be repaired instead of simply spinning up a new host.

Yeah I'm not criticizing it as a charity, just pointing out this definitely isn't "superior to most commercial services."

That being said, removing dependence on single hardware nodes isn't something you need a big team for. I've done failover at 1-person startups.

The post-mortem is honest, but the infrastructure is well below what I'd expect from commercial services.

If a commercial provider told me they're dependent on a single physical server, with no real path or plans to fail over to another server if they need to, I would consider it extremely negligent.

It's fine to not use big cloud providers, but frankly it's pretty incompetent to not have the ability to quickly deploy to a new server.

Did you make Marblism with Marblism?

How about I look at actual unions in software, like the NYT tech union that immediately started undermining merit, making illegal demands, and discouraging high performance.

Every actual tech union that exists is a great advertisement for not unionizing.

There are some unionized tech workers.

I would never want their jobs over mine.

Embeddings are actually generally not that effective for code.

Yet professional athletes don't negotiate the same contract for everyone (precisely why I hate unions and what this union is demanding).

> Things like making sure that AI isn't used in certain capacities to reduce the staff size wouldn't be unreasonable.

Yes it would. Explicitly demanding inefficiency is exactly why unions are terrible for innovation and progress.

Exactly why I would run far away from any workplace that unionized.

Commoditized doesn't mean 0 capex. Literal commodities can in fact be very capital intensive (ex. offshore oil rigs).

In this case, you need capital to stockpile the GPUs.

Yeah I agree those so-called startups deserve your scorn.

Asking employees to participate in a reasonable on-call rotation is not "wage theft." Your antagonistic/adversarial relationship is exactly what destroys cultures.

If you attitude to an outage on a weekend is to ignore it and say that's the company's problem you should simply never join a small startup. You don't deserve equity if you refuse to share in responsibility.

Tripling the cost of running a tech company in Australia is effectively outlawing startups.

This is preposterous and basically kills startups completely.

You genuinely think a seed stage startup should hire a complete "night shift" to ensure no engineers ever get paged "after hours?" This simply kills startups completely.

Customers who insist on on prem require the most support burden and hassles.

Have you shipped commercial on-prem software? Nobody wants to do it for <$1k. It's miserable.

The unfortunate truth (that I've had to learn the hard way as a founder) is that whole IT approval process is inevitable for any vendor indexing source code.

Even when I've tried to offer lower price tiers, ~every company of 20+ people puts a ton of bureaucracy in place before disclosing source code. Dealing with this bureaucracy has a fixed cost, so it's inevitable that many companies end up excluding the lower end entirely.

Last time I got a quote it was >$5k minimum to use code search.

This is exactly the kind of cringe-inducing content that I'd expect from a "PM career coach."

You can communicate understanding of the business without falling into BS slop.

Truly senior people have made such significant and real contributions that (a) their resume is mostly an afterthought and (b) the story

Foundation models aren't primarily trained on ASTs, so you're typically going to have worse results than just using text unless you do extensive fine-tuning yourself.

ASTs also generally don't actually have magical information in them. They won't solve the lifetime issues for you.

Line by line is infeasible, which is precisely why you need to use AI to make larger semantic inferences.

You also don't have to one-shot translate everything. One of the valuable things about the Rust compiler is it gives lots of specific information that you can feed back into an LLM to iterate.

I've been working on similar problems for my startup (grit.io) and think C -> Rust is definitely tractable in the near term. Definitely not easy but certainly solvable.

> Nat Friedman and Daniel Gross own a cluster of GPUs that they rent back to their startups for below market rates.

Not just their startups btw.