---
date: 2025-07-24
url: https://hackmd.io/IG52KQb0T0y368vc0N4cIA
---

# 2025-07-24

People: Jack, Pete LeVasseur, Josh, Jon Bauman, TC, Niko, Tomas Sedovic, Ernest Kissiedu

Notes: Tomas

## Agenda

* Updates and thought sharing
* Project goal

## Updates

### nikomatsakis

#### project goal 2025h2

I proposed a project goal for 2025h2 that amounts to "gather enough data to do our job".

#### thoughts on script

Talked to Jack in prep for Rust Conf keynote. Our discussions yielded a few insights. To start, ultimately, the goal of the doc is to help the Rust org answer questions like

* What is Rust's mission
* What should we focus on for the next 2 or 3 years
* As we do our work, what should we make sure to prioritize / not break / respect

...which we (or at least I) think could would be helped by having answers for "research questions" *we* are most interested in were something like this:

* What makes you {happy, meh, miserable} when you use Rust?
* What makes people choose to use Rust or choose not to use Rust?
* What changes would make you love Rust more, and what change would make move away from Rust?

...segmented by various factors (domain, level of Rust experience, etc). And which then in turn might be *solicited* by a question script that's somewhat different than what I at least was doing, more like this (but letting it go off-script as desired)

* Tell me about the last time Rust let you down?
* Tell me about the last time you were really frustated by something about Rust (the language) or Rust the ecosystem?
* Tell me about the last time you were delighted by something in Rust
* What changes has Rust made that really made an impact on you?
* Tell me about how you decided to use Rust at your job
* Tell me about a time you thought Rust could be good but decided not to use it
* Tell me about a time you thought Rust would definitely not be good and didn't even consider it

#### things I plan to do

* read over all the interviews done so far with the above in mind (I had hoped to do that before this meeting but failed)
* attend Boston area meetups for Swift (today! virtual!) and Python (Sunday) and see if I can get contacts of people who'd like to talk to me about Rust, either because they've tried it or not -- I didn't really prepare a script
* aggressively conduct interviews over next 2 weeks to try and gather a solid base of data
* I'm debating about how to focus that
* I was planning to reach out to folks at Amazon, Discord, and the meetups above at minimum
* And pull some samples from survey

#### slicing and dicing survey

Before going on vacation I had some fun writing up a tool to analyze survey data. It's kind of a "kapiche alternative", it extracts verbatims, etc. I was mostly curious, it didn't yield anything yet, but I might play with it some more. It could also be extended I imagine to handling transcripts. It works using a combination of AI and good ol' fashioned keyword searches and the rest. May go nowhere.

### Pete

Do we have enough diversity in:
* survey responses
* interviews
? (open question to group)

#### things I plan to do

I am shaming myself into now getting Kapiche to work (and I also put it on my todo app right before this meeting)

Continue to interview:
* folks in safety-critical industries who are 1. either using Rust or 2. would like to use Rust
* Codethink, Vector, Elektrobit candidates also getting lined up, need to confirm
* folks from a C and/or C++ background

Some of these transcripts / notes I've been taking should make their way into Kapiche, I should give this a try

### ...

## Notes

Niko: I'm curious Pete to hear more about the safety critical interviews you'e done. Any surprises?

Pete: I talked maybe to five people. Automotive, safety critical. One interesting point from Kai who works at Bosh (their OSS person), his background is cloud and Java. He was told: you have to start investigating in-vehicle technologies. Deciding between C, C++ and Rust. Made sense to pick Rust.

Josh: A lot of people here will be familiar with the environment (e.g. Pete you'll be familiar with safety critical). What things that you've encountered you think would be a surprise to the broader Project?

Jack: I haven't come away with any real surprises. There are things we haven't thought to priorinize, but nothing where there's like "wow, I never thought about that".

Josh: I think that should be a leading item of any document we want to produce. What things we'd be surprised about for people trying Rust. Even if there's nothing or priorities we want to highlight.

Jack: I agree.

Pete: An architect at Toyota -- they have hundreds of engineers working on Rust. They struggle with how to have a cohesive project layout. In automotive people don't trust software engineers. So there are boxes that engineers are allowed in. In C++ they have a project template that everyone has to use that sets things up that makes sense and stops them from doing unwanted things. He tried to do this in Rust, using cargo generate. He's struggling with having a template and being able to evolve it overtime. There may be 50-200 services generated using this that may need to be updated at some point (to switch middleware). Doing Rust projects at scale.

Jack: When you say template, do you mean a set of crates, structure of a project?

Pete: Structure of the project, tools you're using. There's cargo generate that if you point at a folder with a lot of files stubbed out, it'll generate a project based on that. He's using that. But then in the future how do you roll a change to the template to all these projects?

Jack: Anything else that may be surprising?

Jack: We have a talk in September at RustConf. We really only have ~2 weeks to do interviews to get everything in the row. We're interested in thoughts of how to structure the talk. One interesting way is to start by motivating what is the role of the doc. We want the various teams have a basis for their decision. And we hope the doc would provide that base.

Jack: So we'd take a couple of RFCs, highlight what's the main conflict/division that's preventing the RFC to merge. But that may be difficult to base the whole talk on that because you need to find specific data for this to be interesting.

Pete: Something I'm trying to bring to the team is: "we're trying to talk to people we don't normally talk to". E.g. Safety Critical means an influx of people in who bring their own specific complaints. Highlighted the domains we've intervied people from, what they brought up.

Jack: That's a good idea. But it would also be interesting if they said something different than what we already know.

Pete: I have a list of things that came out.

Ernest: What I'd like to see is more of the growing Rust global diaspora. Metrics, data, information based on the interviews. "Here's where Rust is growing where you didn't know Rust was being used before". I've got hackmd to share regarding global comunities.

https://hackmd.io/bzwCLLGnTDSp4W1tRbxpmw

Niko: Ernest, if we can make this stuff happen in the next two weeks we should absolutely do this.

Ernest: I've primed this. The folks from the global jurisdiction are primed. We can startt with the folks at Oxidar (the Spanish-speaking people in the world).

Jack: What do you need from us?

Ernest: Once you've read the document, the question will be answered.

(Ernest quickly walked us through the document)

Jack: I think we should definitely get something scheduled if we can. If we can't find a second person to join one of myself or Niko will join. We could also ask someone else in the project. We should chat more about it on Zulip and set something up.

Jack: Going back to RustConf and the talk there. The conclusion Niko and I made yesterday is: some of it is going to depend on the data coming out over the course of the next week. But we're interested in your thoughts.

Pete: Would something like people trying to use the capiche tool, trying their own interviews, looknig at the result.

Jack: If you had made an account with the tool please share the email with me and I'll try to add you. We'll definitely need to do some actual coding here. I don't know how to do the coding for the interview part. Multiple people need to go through the survey, look at the interviews and find what's the same/different. The more coding we can do the better. Need to find ways to do the coding to get good insights.

Jack: I planned to read through the interview transcripts, write down key bits of information and then look at them in the wider data. At the AllHands we did a scissor section where we took some interviews and cut them up. If we could do that virtually, that would be great but I don't know how to do that.

Jack: But writing down key points you can go back to and reference would be great.

Pete: I can try ot tackle that. Do you have a timeline?

Jack: Our talk is this September, Niko's travelling for two weeks. So I'd say in the next two weeks.

Jack: There are two meetups I'll go to during the weekend. And I'll try to connect to some scientific interviews.

Ernest: I came back from the ?? python meeting in Prague. There was a Rust track that I spoke up. There's a lot of repeated perspectives I got from Pythonistas and their perception of Rust. There's a lot of good will towards Rust. Especially the tooling. There's a lot of love for uv and ruff and PyO3.

Ernest: There was also a C summit where WASM was talked about a lot. I can do a write-up.

Jack: Yes please do a write-up. It would be great for us talking to other people. I'm not all that surprised but this is great to hear.