text stringlengths 0 2.35k |
|---|
**Matan Peled:** It sounds very thoughtful. |
**Mat Ryer:** Yeah, it is. Well, it makes some better programmers, really. I don't know why -- do they teach that at university? Do they teach rubber ducking? |
**Matan Peled:** I teach it when I TA students, and sometimes they have to help them with their homework. Basically, I have them just explain the problem to me, and through the process of them explaining it, they understand what the problem is. And then I tell them, "Well, the next time you should try that with a rubbe... |
**Mat Ryer:** \[laughs\] Just leave it there. Don't elaborate. |
**Natalie Pistunovich:** Just in your reception hours. "Here's a rubber duck. Speak to it." |
**Mat Ryer:** Good idea. You could just have that in your office. |
**Natalie Pistunovich:** A weird thing I learned about rubber ducks is that if you actually put them in a place with water, like a bath, you want them not to have the hole that allows them to squeak, because this is how they get moldy. |
**Mat Ryer:** Oh, really? Why? Because water goes in? |
**Natalie Pistunovich:** Yeah. And it's not very well-ventilated. So if you do plan to bring it into a wet room, then make sure that it does not squeak. |
**Mat Ryer:** So if you're a programmer and things aren't going well and you're trying to rubber ducking and you're crying a lot, keep the tears away from the duck... Because it could get moldy. |
**Natalie Pistunovich:** This would be step ten. |
**Mat Ryer:** Yeah. |
**Natalie Pistunovich:** I could think of a few other things you should do before that... But it can be on the extended list of things you might want to do. |
**Mat Ryer:** Yeah, good point. Tissues, hair dryers... Loads of ways to -- |
**Natalie Pistunovich:** \[laughs\] |
**Matan Peled:** Taking a walk. |
**Mat Ryer:** Oh, yeah. That is a good hack, actually; if you're stuck on something, go for a walk, or think about something else. How many times do you hear people say, just in the shower, or even like asleep sometimes, "I was able to solve this problem." Taking a break is important, and that's really counter-intuitiv... |
**Natalie Pistunovich:** Yeah. |
**Matan Peled:** Yeah. But you feel pressured somehow, usually... And that pressure prevents you from thinking clear. And taking a walk is usually a bit hard, because you need to force yourself to stop thinking of it, and focus on meditation, or walking, or nature... But then your mind has a chance to process whatever ... |
**Natalie Pistunovich:** I'm trying to see what's a good way of tying that to the subject today, but I have yet to find a good one... So in a less smooth transition, no jokes - Matan, you are a Ph.D. candidate at The Technion in Israel, and you're researching meta programming and static analysis. You have worked in all... |
**Matan Peled:** So for my bachelor's degree I did computer engineering, because I had this (let's call it) romantic idea that I should understand how a computer works, from physics to software... |
**Mat Ryer:** Wow. |
**Matan Peled:** It was very cool, and I learned a lot, and I like my degree a lot, but it made it tougher than it might otherwise would have been, because I'm a programmer, I did software engineering before I started studying, so I was already good at that, and adding something that was completely new and different li... |
\[08:09\] So I went back for my master's, and finished that, and now I'm in my Ph.D. And I like it. I don't know if I'm going to do this forever, but for now, this is fun... And I get to play in my sandbox, and make sand castles, and toy around with my own toy ideas, which is what I like about academics. |
**Natalie Pistunovich:** You mentioned code monkey, and I have to make the joke now, because I still feel overdue from the less than smooth transition from before... Have you tried being a code gopher? |
**Mat Ryer:** \[laughs\] |
**Natalie Pistunovich:** Fun. So what was your master's research topic? |
**Matan Peled:** So my master's research was making a programming language. I was in programming language design. What my programming language did was it took slideshows, and I wanted to add animations and I wanted to basically make a language that describes what the animation is, what the motion on the screen is. |
I liked it, it was fun working on it, but let's just say that its main purpose was to get me a master's degree, and it didn't turn out to be especially usable. It had a couple of really cool ideas... For example, the idea was that I could build a big animation from basic parts, and that if I wanted to make a change in ... |
So I had a few cute ideas in there, of making it like a physical system of springs, where you can add a spring and everything basically reacts based on the change you've made. And it worked, and it was nice, but it was not a production system, let's put it that way. |
**Natalie Pistunovich:** Because it had no tests. |
**Matan Peled:** Actually, it did have tests. It's just that basically you had to think like me in order to use it... And think like me at that very specific time of me writing it. |
**Mat Ryer:** So how important is that to you then? Because you talk about you like being in the sandbox, building sand castles... How much of that is anchored back to something that could be practical or useful? |
**Matan Peled:** Well, that's a good question. So my thinking is that I want it to be anchored back to something that is practical and useful... But I'm okay if it's not. I'm okay going often to the wilderness, and exploring, and finding out things... And if it turns out that they're not impactful, and they don't make ... |
**Mat Ryer:** Kind of like a startup attitude, really... Being able to fail, and having that freedom to fail is quite important. It gives you that extra permission almost to do things that otherwise people might not have a chance to do. |
**Matan Peled:** Yeah. In one sense, I think startup culture is about being small and agile, and being able to push yourself into a niche that a bigger company just wouldn't fit into, because they can't be flexible enough just to think about it that way, to allocate research for that thing... But on the other hand, I t... |
**Natalie Pistunovich:** \[12:32\] That sounds a lot like blogging about fun projects that you have. |
**Matan Peled:** It's blogging on steroids. It's blogging with a lot of formality added. And that's why science communication, where scientists blog - that's a thing that's happening. Academics usually love to write; that's what they do in their day-to-day, they're usually good at it. They make blogs. Academic Twitter,... |
**Mat Ryer:** Oh, I bet that's amazing. Is it good? |
**Matan Peled:** Yeah, if you're interested in that sort of thing. |
**Natalie Pistunovich:** Have you been tweeting about static analysis tools? Or static analysis in general. Or wait, maybe we can start with saying what is static analysis... |
**Matan Peled:** So static analysis - basically, what we want to do is figure out certain properties about code. A property could be, at its most basic thing, "Does this program have a bug?" or "Does this program succeed?" And we want to do all that statically, and that means without actually running the program, becau... |
So static analysis can be anything from where this function is called, to "Is this program written in the correct style?", which is something that's not that big of a problem with Go, because we have go fmt, which is supposedly also like a static analysis tool. But other languages have things like linters, which tell y... |
But static analysis can also be a part of refactoring where you want to rename a method and you want the IDE to find out where all the calls to this method are, and use static analysis to find that. |
**Natalie Pistunovich:** Yeah, right-click refactor. |
**Matan Peled:** Exactly. |
**Break:** \[14:44\] |
**Mat Ryer:** Static analysis must be easier on typed languages that are strictly typed, versus languages that are dynamic, that have heavy runtime elements... Is that true? |
**Matan Peled:** Oh yes, definitely, because type checking is also a type of static analysis. Basically, by having types, you're giving the tool a lot more information than it can use. And if it has more information, then it can do more stuff. |
One of the basic truths of computer science is that static analysis is impossible. You have the halting problem, which Alan Turing proved way back when, which says that basically you can't make a program that says if another program will halt. And the proof of that is very cool, because basically he said that "Well, if... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.