text
stringlengths
0
1.49k
**Erik St. Martin:** ...which I've done. \[laughter\]
**Michael Stapelberg:** Me too, me too.
**Erik St. Martin:** And I will admit, I am not the best C programmer, nor embedded systems person, so...
**Brian Ketelsen:** Better than me.
**Erik St. Martin:** ...weird things happen. And if you want proof -- so Brian was cooking a whole pig one weekend and we literally just threw together something like this, like you said, kind of in a weekend and whatnot. And you can see proof of this through the way I did the serial connection between the Wi-Fi chip a...
**Brian Ketelsen:** Not quite right.
**Erik St. Martin:** This would have been so much nicer if I could just use a good library for this and I wasn't trying to cobble it together really quickly in a night.
**Brian Ketelsen:** Going back to i3 for a moment - I hate to do this, but after this show, Michael, I need to send me your mailing address, because I need to send you a bill for anti-depressants. It just occurred to me that the whole reason I can't stand computers anymore is because of i3. I only want to develop when ...
**Erik St. Martin:** I think you should invert that... I think that you should have to pay him for not having to take them during the periods of time that you used i3.
**Brian Ketelsen:** \[laughs\] There you go. So that brings up a good point - I contribute monthly to several [Patreon](https://www.patreon.com/) accounts for open source projects that I truly love... Does i3 have a Patreon?
**Michael Stapelberg:** No, we do not currently have one. With a project where there's largely a single maintainer, but then so many contributors, it often becomes an issue of fairness, like "How do we distribute the funds?" If people approach me and say they really wanna donate, I usually accept their donations, which...
We're certainly not trying to cover any sort of development costs or make a living off of i3. It is a spare time project and it will remain a spare time project.
**Brian Ketelsen:** It's a labor of love. So speaking of the i3 sticker, you'll get a kick out of this - my mother came and visited two weeks ago. The day before she left, she looked into my office and she said "Are you using all of those laptops? Are there any spares that you can send home? Because my laptop is dead."...
**Michael Stapelberg:** \[36:23\] Oh, absolutely. Yeah, absolutely.
**Erik St. Martin:** So you talk about all these projects that you're doing on the side - what are you working on for day job? What does pay the bills?
**Michael Stapelberg:** Currently, I'm lucky to be employed by Google in the European headquarters. In fact, I am working on Go in a capacity; not as part of the Go team itself, but we're working in this so-called Frameworks Go Team. That's a team where we wanna make it easier for other teams to develop software that r...
So we're trying to make it easy for them to run good Go code, and of course, Google has many years of experience in C++ and Java for running all sorts of applications, but in Go it's not quite there yet, so that's what I'm working on.
**Erik St. Martin:** Oh, that's awesome. So you mostly work with the internal teams...
**Michael Stapelberg:** Yeah, absolutely. Very occasionally, there are places where I cannot only contribute to something internal, but also something external, and then I happily do that. But the large part of it is focused on the internal stuff, yeah.
**Brian Ketelsen:** I read a really interesting article this morning, talking about Google's internal software infrastructure versus open source, and the article suggested that the software, the frameworks that Google uses internally are five to eight years ahead of open source equivalents... And if you wanna have a se...
**Michael Stapelberg:** I wouldn't be able to say exactly how many years it is, but yes, I definitely got the feeling when I joined that this is years ahead of everything that I've known up to that point. In fact, I would love it if open source would look more like what we have internally in a couple of years, just bec...
As just one example, a different project of mine that I run is [Debian Code Search](https://codesearch.debian.net/), which is a regular expression code search engine for all of the open source software that is found in Debian. When I launched it, we had quite some trouble getting the resources for it, because you need ...
**Carlisia Thompson:** Is Google going to open source a lot of the infrastructure or some of the infrastructure?
**Michael Stapelberg:** \[39:56\] I think that's definitely the trend. Infrastructure is always hard to really open source, as in infrastructure software and systems. But if you read about it, there was the release of [Abseil](https://abseil.io/) recently, which is sort of Google's base libraries for C++ and also for P...
I'm very excited about this release, because it means that other parts of Google will also be released. Because essentially, every project that we have uses the base libraries; that's why they're called "base libraries." If the base libraries are not released, you need to jump through so many hoops whenever you want to...
**Carlisia Thompson:** The reason I even know this is because [JBD](https://twitter.com/rakyll) is leaving the Go language team to work on the project to open source some part of the infrastructure, and I was wondering if you would know more specifically what that is.
**Michael Stapelberg:** Unfortunately, I didn't even know that she was moving, so... I don't know, sorry.
**Carlisia Thompson:** Yeah, more like what the plans are for open sourcing... What parts of the infrastructure is what I'm curious.
**Michael Stapelberg:** Right. I think that is a decision that the individual teams actually need to make for themselves. I think in general the climate is pro-open source and people are encouraged to think about it and consider open sourcing it, but ultimately there's no top-down mandate to open source everything. Tha...
I think if the team in question has the man power to do it and has the will to do it, they will now be much more empowered than they used to be.
**Carlisia Thompson:** Cool.
**Erik St. Martin:** That's awesome. And even just things like [Kubernetes](https://kubernetes.io/), right? That's not a direct open source release of an infrastructure tool of Google, but it's the recreation of one that's not so tied to Google's infrastructure, which is really awesome that we get to share in these thi...
**Michael Stapelberg:** Yeah, absolutely. And in fact, Kubernetes was one of the things that I was so happy about when it was released, because I was like "Finally, there will be products like this, which offer Kubernetes as a service." That's really what I think will be very useful, because -- for example, for i3 we u...
We run part of it on [App Engine](https://cloud.google.com/appengine/). For example the bot that services GitHub issues runs on app engine. It would be entirely conceivable that we could run more of these services on like a public or semi-public hosted Kubernetes, like the [Google Container Engine](https://cloud.google...
I'm excited also for more service providers to step into that market. I think Microsoft in fact has an offering on Azure; you probably can just get a couple containers in Kubernetes... And I hope that this becomes much more of a commodity. I don't wanna manage all of these machines, much like in the gokrazy project; th...
**Erik St. Martin:** Yeah, just take my stuff and put it somewhere, please...
**Michael Stapelberg:** Exactly.
**Carlisia Thompson:** So the [expanderr](https://github.com/stapelberg/expanderr) project - obviously, super cool! For people who don't know, it's an automated way to add the `if error` statements. The tool automatically recognizes if your function returns an error, or the function you're calling returns an error, and...
**Michael Stapelberg:** Yes.
**Carlisia Thompson:** Do you know of the other editors having plans to add it? Because I want it.
**Michael Stapelberg:** \[44:07\] Sure, yeah. To answer the direct question, and then maybe I can talk about how the project came to be... There is in fact a pull request by someone who is currently adding it to the Vim Go plugin; that all needs to be merged. I think [Fatih](https://twitter.com/fatih) is very busy with...
I think we briefly brought up adding it to Visual Studio Code as well, and they were very receptive of the idea. It will be added to more editors as people make that little bit of effort and just integrate it.
**Carlisia Thompson:** That'd be awesome!
**Brian Ketelsen:** That's awesome!
**Carlisia Thompson:** Thank you for that tool. I mean, all the tools you do.
**Erik St. Martin:** That seems to be one of the common complaints too from people coming into Go, too... They feel it's super verbose to have to constantly do that, so I think that's gonna be an easy solve for a lot of people. "I hate typing the error checks!", but you don't have to.
**Michael Stapelberg:** Exactly. That was actually exactly the feeling that I had. I was writing a Go program, and I don't really mind typing these error checks whenever you're modifying an existing program, because they tend to not be so pervasive. But when you're prototyping something, especially something that inter...
So I found myself typing all of these error checks over and over again, and I know that, for example, in the blog post by [Russ Cox](https://twitter.com/_rsc) at the beginning of the year he said that he will want to look into improving the error checking situation... So I figured, "Well, it's good that they're working...
So at some point - in fact, shortly before GopherCon - I was considering doing this, and then almost as a joke, I started and figured "Well, could you do this? Could you write a little tool that when you invoke it, it just expands whatever is under the cursor?", and it turns out it is possible.
Then we had a Go meetup in Zurich. I wasn't still quite sure about the idea, because error-checking is such a hot topic in Go, and I wasn't quite sure whether other people would like it... So when we had the Go meetup in Zurich, there were a couple of presentations, and then after the regularly scheduled presentations ...
**Brian Ketelsen:** \[47:47\] It must be alright. Yeah, it's fantastic. I love how smart it is. I'm watching the little video on the page over and over, and it's aware enough to return the right things when you change your function signature - that's brilliant, I love it.