wlamartin · HackerTrans

11 месяцев назадdiscuss

I don't think people in this thread aren't really confused about MCP. They are confused that you claimed, or at least insinuated that an LLM might skip the schema validation portion of an MCP tool call request/response, which was originally demonstrated via Claude Code. Hopefully you can understand why everyone seems so confused, since that claim doesn't make any sense when the LLM doesn't really have anything to do with schema validation at all.

wlamartin

в прошлом годуdiscuss

I'm not sure whether you're confused, or I'm just having a horrible time understanding your point. The MCP server really does just serve requests with responses via a mechanism that satisfies the MCP spec. The MCP hosts (e.g. VSCode) work with an LLM to determine which of those tools to call, and ideally work with users via confirmation prompts to ensure the user really wants those things to happen.

What am I missing?

I do believe there's more that the MCP Server could be offering to protect users, but that seems like a separate point.

wlamartin

2 года назадdiscuss

I've been using your app for the last 3-4 months very successfully. There are a few niggles here and there but overall it's been exactly what I needed, and I'm very grateful for it!

wlamartin

2 года назадdiscuss

Just as a note, the GitHub CLI doesn't use bubbletea itself right now, though it does use other charm libraries such as lipgloss and glamour. That said, it's quite likely that at some point we will use huh for our prompting library, which does use bubbletea.

wlamartin

2 года назадdiscuss

Ah, I think you might be pleasantly surprised that this is an area being focused on right now with attestations[1] for example, here are the attestations for the GitHub CLI[2].

1: https://github.blog/2024-05-02-introducing-artifact-attestat...

2: https://github.com/cli/cli/attestations

wlamartin

3 года назадdiscuss

Recently I've been wondering if there is a "build your own X" for some of these concepts. For example, there is https://github.com/xyproto/vt100 which seems relatively straightforward (though maybe not "simple") to learn from but are there any resources that would actually teach this stuff?

wlamartin

3 года назадdiscuss

I'm a maintainer for the GitHub CLI. When our prompting dependency became unmaintained, the folks from Charm reached out to us looking to collaborate on a replacement (collaborate is a strong word for the amount of work they actually did compared to us).

While we haven't yet been able to prioritise using this package, our discussions with them were great and I think that in typical Charm fashion, they've created something really awesome here. In particular, we mentioned to them our desire to have accessibility as a first class experience and I think they really took that on board.

wlamartin

3 года назадdiscuss

Recently, I've been toying with including asciinema in automated tests. Spinning up a TUI or richer CLI experience, sending it keystrokes, asserting on the output, all wrapped up in asciinema so that if it fails I can see what went wrong via the playback.

Jury is still out but it's been an interesting experiment.

wlamartin

3 года назадdiscuss

When recording finishes you are given the option to upload or save to local file.

wlamartin

3 года назадdiscuss

Northern Irish, living in the Netherlands, commenting on hacker news, there must be severals of us!

wlamartin

3 года назадdiscuss

Years ago when I was starting to interview for Uber in Europe I went through some data structure exercise which I failed at horribly. When I asked the hiring manager whether this was representative of the job, because I would probably not be good at it they said "no", and to follow up, when I asked why we did it they said, "in the next round my colleagues in the USA are going to ask you something similar".

I guess this wasn't entirely without sense but sort of seemed to be missing the point.

wlamartin

4 года назадdiscuss

https://relatedwords.org/ is also pretty neat for this. It's slightly less strict than a thesaurus so it can aid in exploring the semantic space a bit more.

wlamartin

4 года назадdiscuss

I don't follow this. My assumption for the repository interface is something like (language, error and domain agnostic):

   interface Repository<T> {
      FetchAll() T[]
      Fetch(id) T
      Persist(T) id
   }

Why would the SQL statements be reflected in the behaviour that the interface is providing?

wlamartin

4 года назадdiscuss

Such a thing exists: https://github.com/BurntSushi/go-sumtype

wlamartin

4 года назадdiscuss

  Location: Amsterdam, The Netherlands
  Remote: Yes
  Willing to relocate: No
  Technologies: Go, TypeScript, Rust, Terraform, Kubernetes, Containers, FP
  Résumé/CV: https://www.linkedin.com/in/william-martin-41400517/
  Email: william[dot]la[dot]martin[mailsign]gmail[dot]com

Things I enjoy:

  Mentoring
  Pair Programming
  Test Driven Development
  Systems Engineering
  Team Psychological Safety
  Balanced Teams
  Feedback

Looking for a Senior/Staff level Individual Contributor role. I'd enjoy a job using a language with great domain modelling support (e.g. F#, Rust, Elm) to continue learning how to do that well. I'd enjoy a job where I can continue learning from others directly, as well as mentor others directly.

wlamartin

4 года назадdiscuss

I find "A Secret Sleeping in the Deep Sea" really takes me back to my childhood.

wlamartin

4 года назадdiscuss

Sure, there's many ways to skin a cat. Given the option, I wouldn't choose either of these, and instead use something like filterMap[1] which I think conveys intent better than a fold or flatMap.

Btw, I'm assuming that the original map questioner wasn't solely using flatMap for side-effectful iteration, which reading again, I'm a bit suspicious about.

1: https://gcanti.github.io/fp-ts/modules/Array.ts.html#filterm...

wlamartin

4 года назадdiscuss

They don't want to keep the unmapped value in the resulting collection at all. flatMap allows for removal of an element in one traverse, unlike filter+map with eager behaviour.

wlamartin

4 года назадdiscuss

tRPC is a good choice for TypeScript: https://trpc.io/

wlamartin

5 лет назадdiscuss

> The only thing that's different is that you get a single value you have to check for errors rather than a separate value..

The type system confines you to a set of reasonable cases that allow a caller to reason about the state of the program. This has two benefits for the caller:

1. It is required that the caller check whether a return value is success or failure in order to access the value they want. There is no possibility to mistake one case for another.

2. In the space of valid return values for idiomatic Go function signatures, 50% of them are unidiomatic and end up being ignored. It is far clearer for a caller to understand what is expected of them when valid values exactly overlap with the space. It is far clearer for an author to convey expectations to a caller for the same reason.

Now I must admit, good conventions and tooling in Go account for the vast majority of cases and I don't personally mind that much, but that's a different conversation than API design.

comments