text
stringlengths
0
2.35k
**Akshay Shah:** The way it's done here, yeah.
**Jon Calhoun:** We've introduced you by saying that if you've used Zap, you've probably used your code, and now you're telling us that that was a bad idea.
**Johnny Boursiquot:** \[laughs\]
**Akshay Shah:** You know, for a bunch of reasons... I was at Uber at the time, I was writing a service mesh thing for a Stubby-like protocol... It had tight performance requirements... And we were feeling really good. We had written this service mesh proxy thing, we'd met all the perf requirements... And Johnny, you'r...
**Johnny Boursiquot:** \[laughs\]
**Akshay Shah:** And you add a couple of log statements and a couple of metric increments, and all of a sudden I'm staring at 1,000 heap allocations. The perf budget went out the window, and you're like "Well, we could just ship it without the logs or the metrics, but that's not good..."
And so to fit with the log infrastructure of the company, we had to emit JSON. And there was really no facility for running a regex over a string in the log ingestion and dashboarding stuff. And so I kind of invented this ultimately very fussy API for producing logs, just to avoid heap allocations, more or less, and to...
**Johnny Boursiquot:** And you thought you'd unleash it on the rest of the community... \[laughs\]
**Akshay Shah:** No -- so actually, internally at Uber what we did is we just put this thing behind a facade that looks like the old logging library... And it was like 30% faster, and we're like If everything just got 30% faster for no code changes, this is a huge win." Right? There's zillions of cores just parsing map...
**Johnny Boursiquot:** Yeah, we just like to give money to the cloud vendors. That's cool. Keeping that whole industry in business.
**Jon Calhoun:** Johnny is not biased at all here...
**Akshay Shah:** It was just - for the amount of effort spent golfing allocations out of this one log API, the fact that end-to-end this is just stupendously inefficient... it should matter. We should pick a format that, at the very least, if you want this really fussy logging API, in between your program and the ultim...
**Johnny Boursiquot:** Nice. \[laughs\] Oh, we're gonna see how well that one does...
**Jon Calhoun:** I'm just gonna be quiet, because I don't -- the upside to being a one-person company is that I don't really work on things that need to worry about any of those performance issues or anything that, so I can kind of get away with whatever I want.
**Akshay Shah:** I used the standard lib's log package on all of my personal stuff.
**Jon Calhoun:** I use println() a lot for printing stuff out, like if I need to know something... But that's because it's just me, so it's like, "Okay, this is pretty easy to deal with." I don't have a crazy amount of traffic. I'm currently not running any services with more than 100,000 users, so it's not too crazy.
**Akshay Shah:** That's a very successful software business.
**Johnny Boursiquot:** Seriously.
**Akshay Shah:** 100,000 people is a lot.
**Jon Calhoun:** It's not concurrent, for sure. It's like, there's 100,000 signups for a free course, and at this point, probably half of them haven't logged on in who knows how long.
**Akshay Shah:** \[01:04:11.25\] How are you not a multibillion-dollar VC-funded juggernaut at this point?
**Jon Calhoun:** Because that's a free course. When things are free...
**Akshay Shah:** Got it.
**Johnny Boursiquot:** \[laughs\] Well --
**Akshay Shah:** That hasn't stopped anybody for the last five or six years...
**Johnny Boursiquot:** That's true, that's true...
**Jon Calhoun:** I'd have to have some grand vision then, and I don't have that. My vision is just to try to help people learn Go.
**Johnny Boursiquot:** And just to make it free.
**Akshay Shah:** I'll tell you, I'll send you a slide deck for a 10% cut.
**Johnny Boursiquot:** \[laughs\] Oh, man... That's awesome. That's awesome. So you don't have an unpopular opinion, Jon?
**Jon Calhoun:** I do not. Johnny, what is your unpopular opinion?
**Johnny Boursiquot:** Okay. Well, since you and I both teach Go, so you might appreciate this... I think if you have a training, where you're not actually typing code and showing people how it actually feels and looks like to write the code in an editor... If all you have is slideware, and things, and animations and t...
**Jon Calhoun:** I agree with that 100%. In almost everything you can teach. Just to sort of back that up - when students learn about algorithms in school, I feel like half the reason they don't really understand them is because they're often not seeing code, and they're not using it extensively... So they might see a ...
I did Programming Teams, so I had to code this type of stuff all the time. The programming team is like a Top Coder, or Google Code Jam, that type of stuff, is what we did all the time. I got to work with other people who were very good at breaking it down and showing you "Here's ten examples of actually coding this th...
**Johnny Boursiquot:** Oh yeah, absolutely. The code-along is encouraged. As a matter of fact, this is a good segue into a shameless plug... By the time this episode is available, I have a new training coming out with Lincoln Learning, actually... It's a hands-on introductory Go course. So for those who have been sort ...
So I'm working very hard on it, looking forward to actually putting this thing out in the world... It's kind of like one of your babies, right John? You work on it, you toil, and you sweat, nights, weekends, all the things, and then it comes out and you're super-proud of it... This is definitely one of them for me.
**Akshay Shah:** Is this the thing that's launching tomorrow, that you were mentioning at the beginning?
**Johnny Boursiquot:** CodeSpaces? Yeah, that's the thing launching tomorrow, along with GitHub, at GitHub Universe, and some CodeSpaces stuff. It's upon us, so hopefully I'm not breaking any rules and stuff, and having sort of a recording that shows up a week or two later... So it's just in time.
**Akshay Shah:** Nice.
**Jon Calhoun:** Nice.
**Akshay Shah:** I learned programming, like commercial programming, before my first job, with a Learn Python book, that was very much in line with your opinion, Johnny. It was Zed Shaw's Learn Python the Hard Way.
**Johnny Boursiquot:** \[01:08:15.23\] Yeah...
**Akshay Shah:** And it's literally just a bunch of code that you just hand-type and push Enter, and try and figure out why it did what it did.
**Johnny Boursiquot:** Yup.
**Akshay Shah:** And I remember I burned through the whole thing, because I had somehow convinced them that I knew how to program before I started, but I didn't really...
**Johnny Boursiquot:** \[laughs\] Nice.
**Akshay Shah:** Certainly not in Python. And it was under the gun to really figure it out quick.
**Johnny Boursiquot:** Those are the best, right?
**Akshay Shah:** Well, I showed up there, and I was like "I know how to code", and they kind of looked at me and they're like "Yeah, that checks out. Come on in." It was not the greatest... \[laughs\]
**Johnny Boursiquot:** Nice.