CppCast - STX

Episode Date: August 20, 2020

Rob and Jason are joined by Basit Ayantunde. They first discuss a blog post on plain old data types and Visual Studio's Address Sanitizer support. Then they talk to Basit about the STX library. News No more plain old data AddressSanitizer for Windows: x64 and Debug Build Support Meeting C++ Online Tickets Available Links STX Basit's Patreon Sponsors PVS-Studio. Write #cppcast in the message field on the download page and get one month license PVS-Studio is now in Compiler Explorer! Free PVS-Studio for Students and Teachers

Episode 261 of CppCast with guest Basit Iyantunde recorded August 15th, 2020. Sponsor of this episode of CppCast is the PVS Studio team. The team promotes regular usage of static code analysis and the PVS Studio static analysis tool. In this episode, we discuss plain old data types. Then we talk to Basit Ayyantande. Basit talks to us about the SGX library and other open source work he's done. Welcome to episode 261 of CppCast, the first podcast for C++ developers by C++ developers. I'm your host, Rob Irving, joined by my co-host, Jason Turner. Jason, how are you doing today? I'm all right, except for this cloud hanging over me.

Of smoke, that is. Yeah, what'sado like these days well it's just a bunch of forest fires it's standard you know wildfire season i guess out here so uh it's uh the other day like i couldn't even see the yesterday i couldn't even see the sun at like uh 6 p.m it was so dark with cloud with smoke are they like out of control or are they fairly well controlled by the fire department i don't know i mean they had to evacuate a few areas but it's all you know just kind of normal it's funny because i recently tweeted about this i like you know just the extreme weather like this is just part of living in Colorado. We get blizzards, monsoons, wildfires. We don't get earthquakes.

We do get tornadoes. It's fine. Okay. All right. Well, the top of Arizona likes to read a piece of feedback. This week, I saw this tweet from Patrice Roy, who we've had on the show, I think, three times before. I haven't talked to him in a while. But he wrote this tweet saying,

Not commuting and working from home and working all the time makes my podcast backlog grow bigger by the day. I miss listening to CppCast. It's on my short list of must-listen podcasts. And I saw this follow-up reply from Bjorn Fahler. And he said when he commuted, he listens to a lot of fascinating podcasts, but now there's no commute, so little listening time. Then an odd thing happened. Long lunch walks, bicycle rides, whatever, and podcasts again. Only many aren't airing some summer breaks, which I guess some podcasts take off for the summer.

Not us. Why don't we do that? Maybe we should do that. I did kind of think about this, and that's something that i should really do i also you know don't listen to many podcasts anymore because i don't have that commute into the office which i know you have not had for a long time jason like 11 years yeah yeah but i i should do more walks or bike rides or something because you know that commute time just winds up being more work time or just sitting around the house and i should really be better about that yeah yeah well we'd love to hear your thoughts about the show you can always reach out to us on facebook twitter or email us at feedback at speakass.com and don't forget to leave us a review on itunes or subscribe YouTube. Joining us today is Basit Ayuntande. Basit is an undergraduate at the University of Eloran,

Nigeria, where he majors in mechanical engineering. Basit has four years of experience writing industrial software, both as a contract software developer and intern. He is one of the C++ Now 2020 scholars. He contributes to open source projects like TensorFlow and CMSIS and authors STX and a number of open source C++ libraries. Baza, welcome to the show. Thanks for having me. So I've been curious about this because I saw on your bio, yeah, that you, a C++ Now 2020 scholar, we didn't talk a whole lot about C++ Now because, well, it was canceled this year, right? So were you planning to come to Aspen?

How did that all work out? Yeah, I was planning to come but the conference was cancelled. I was even about preparing for my visa and stuff. It was cancelled so I disrupted my plans. That's unfortunate.

I'm kind of curious, you're majoring in mechanical engineering but you've done so much software development work. Are you learning C++ as part of your curriculum in mechanical engineering, or is that just something you became interested in on your own? I actually became interested in it on my own during my first year. That was when I started learning C++. Very cool.

The last engineering program that I was really familiar with, everyone had to learn Fortran in their first year, which just makes me sound old. I don't know if that's still true or not. Actually, I did some C programming in my second year, but it wasn't really intensive. Right. All right, well, Basit,

we have a couple news articles to discuss. Feel free to comment on any of these and then we'll start talking more about your STX library and some of the other open source work you've done, okay? Alright, so the first thing we have is this blog post titled

No More Plain Old Data. And it's all about plain old data or pod types and uh how the you know the feature to you know do the type trait to detect a pod type is actually being deprecated in c++20 which is i don't think something we've talked about before but there are now the type traits for is trivial and is standard layout uh and you know the concept of pod types are kind of going away i guess but you can think of a pod type as both a trivial and standard layout type,

and that's what this blog post is basically going over. Lots of good examples in this one. I found it pretty helpful, actually, because when I saw that the is pod was being deprecated in C++20, I'm like, really? Why do they take the nice things away from us? But this article gives enough background for it to really

make more sense. Yeah, definitely. Do you have a chance to read through this one, Basit? Yeah, I'll also read through it. I think it's a very good idea, separating the trivial and POD types, because basically those PODs are more

like trivial types, but it's really very constrained constructors. Next article we have is from the Visual C++ blog. This is about the address sanitizer support in Windows with Visual Studio. We've definitely

talked about this one before, but they keep adding and expanding on the address sanitizer functionality in visual studio and it now supports x64 i believe when it first started it was just for x86 it also has debug build support and this is a really good post if you do want to try out the the visual studio address sanitizer because it goes over how to set it up properly, what some of the known issues are, and just seems like a really helpful post. I know I did try this out when I think it was first announced and I had some trouble, but I'll have to take another try at it now that

they've added x64 and debug and see if I can have more success with it now. This will be a little bit of a mental shift for me because, I mean, for so many years now, I've been using Visual Studio, and I'm like, you know what, I really need to run this through Agile Sanitizer. All the projects that work on our cross-platform anyhow. So I just run it in Linux. You know, I have to remind myself,

wait a minute, I should be able to actually do this in Visual Studio now. I also like this comment. I just want to point out that they said, we have fixed bugs in our tools thanks to customers who were early adopters and building large internal systems like Office. I think running Office with address sanitizer probably counts as an effective test. Yeah, definitely. That's a big product. That's a very big product. Yeah. yeah bazza do you have a chance to use this and are you a visual studio user or a linux user linux okay all right uh and then the

last post we have is meeting c++ 2020 and uh there's now a list of accepted talks and definitely a couple changes uh to the conference this year, which I guess we should expect. They're going to a single track as opposed to, is it normally three or four tracks? There's four tracks over like three days. This looks like a single track over, actually, I don't know how many days it's over. It's 11 talks. So it's definitely not one day. I don't think so.

Yeah. it's 11 talks that's definitely not one day i don't think so yeah yeah uh but single track now and there are going to be options to buy online tickets and i believe the plan is to still have some uh in-person conference goers but it'll probably just be you know local people in berlin uh you know We still have to worry about COVID, even though I think Germany's numbers are fairly low these days, which is good. Yeah, I can't go. At least not at the moment. No, I mean, we're not even allowed to travel there at the moment.

Yeah. Yeah. All right. Basit, we've mentioned a lot of open source work that you've done in your bio. Could you start off by telling us about your STX library? Yeah, sure. So STX is a set of error handling and utility extensions to the C++ standard library. It was primarily intended to address a few shortcomings or features that we feel would be nice to have in C++.

So we have libraries like the monadic optional and error-undue libraries and the backtrace library, which actually make working with C++ easier. Right. I was planning to ask about that because I did notice you had your own optional implementation in there. And I'm like, well, optional isn't C++17. So what makes your optional implementation different? The primary concern was with C++ exceptions, especially if you're in the operating systems on the base system side. So it would be a big drug for you our optional type uses the feel-fast and customizable

abandonment mechanism that we call panic so with my experience in the system software it is much much more desired also on two chains like extensor you don't really even have a standard exceptions mechanism. So it can even be very heavy for some embedded systems platform. Okay. Can you tell us a little bit more about... I'm sorry, go ahead. Okay. With the monadic operations too, you can like... With the error handling logic, you can separate them as in pipelines

instead of writing them manually. So it actually makes even error handling easier than std optional. Yeah, I was recently working with optional myself and thinking that felt like a lot of boilerplate to be like, if this, then do this operation. If the optional's not set, then do this other thing or whatever.

So the monadic operations... Go ahead. Steve Brown wrote a paper about that for sts optional. Right. I haven't really been tracking its progress. Yeah, I know that there's I have no idea what the progress is

on getting that standardized. Me too. Can you tell us more about the panicking? I don't think I've heard of other libraries with... It says runtime panic hooks. Can you tell us a little bit more about that? The runtime panic hooks are like customizable hooks. Hooks are like runtime mechanisms

that used to customize behavior or functions. So you can... If you want the library to like feel fast abort you can set it to abort the panic instead of like causing your program to like immediately terminate it to just abort instead if you want to see to log to disk to log to disk instead before terminating so it gives you because there are some mission-constrained platforms. For example, the Mars rover, if you are launching

into space, you don't necessarily just want it to abort mid-space, mid-mission. You might want it to log to disk so you'll be able to fetch the reasoning for the abort and other error reports. So you can do that with the panic hook. Assuming you have any behavior you want,

once an error occurs, you can do that with the panic hook. Is it like a global handler, like set terminate or whatever, like that kind of thing? Yeah, exactly. Both results and

optional, you report to it. Interesting. Is this library only for C++ 20 or could you use it with earlier versions of the standard? Initially, we had C++ 20, but we had to remove some C++ 20

features like the concepts and some other C++ 20 specs features from it. Most of our users some other C++ 20-space features from it. Because most of our users were using C++ 17. I'm still kind of curious about your panic handler, and I'm curious if it's like thread safe or what happens in like a multi-threading situation if the panic handler is called. Yeah, it's actually trace-safe. We intentionally made sure that it's an atomic operation, so all the interactions with the panic handler are atomic. Neat. Okay. Now, I've got the

documentation up on my screen here, and it looks like you are not afraid to use constexpr. So is constexpr supported throughout the library? It's just for the results and option type non-tragular here. You also mentioned backtracing here. Is that to produce stack traces so you can debug what went wrong in the case of a panic? Yeah, that's for the panic handler. So once you can also customize the panic book to give you a detailed backtrace on the planning purpose. You know, for example, if you have a six segv, you can, especially for beginners, it's very difficult to trace exactly where it's occurred.

So the backtrace library can use it to know where the sigv occurred, which is the reference in null pointer or an invalid pointer. So it's actually very useful for that purpose. So you said you're a Linux user when we were talking earlier. Does the library support Windows by any chance? I have some of our contributors have tested on Windows, so I'm very sure it works. I also have test sites for it on Windows.

Okay. So you just mentioned signal handlers. That sounds like something I don't think can work on Windows. And I just realized at this moment, I've done all kinds of signal handling code on Linux. I've never even concerned myself with the question on Windows. I have no idea. So do you know what that story is?

I know you said your contributors have worked on that, but is there signal handling type code for Windows as well? Yeah, it's cross-platform. Okay. It's usually a serial runtime, so there isn't really a problem with that. Okay. And backtrace, similarly, does it work across platform?

Yeah, both Windows and Linux only. Okay. Okay. I also noticed your library looks like it's got some awesome documentation. It looks pretty clean if you look at the generated documentation you have. And I was curious what you're using to generate your documentation. For your documentation, I'm using Doxygen. Okay.

I've heard some people complain. Oh, sorry. Yeah Doxygen. Okay. I've heard some people complain... Oh, sorry. Yeah, you can go on. I heard some people complain that Doxygen isn't really keeping up with C++17 and C++20, and I haven't used it myself a whole lot lately, so I was curious if you ran into any of those problems. Yeah, it's

not up to date with C++20, but I didn't run into an issue with C++17 support. The only issue I ran into for C++20 was with the concept type. So it treats concepts as variables. It actually gives good documentation, although it's not really adequate source level information. I also tried ClangDoc and various other documentation tools, but they didn't really give

me the right amount of source level documentation as Doxygen. That's interesting. I don't think I've ever heard of ClangDoc. It's still a work in progress. It's not yet complete. Okay. Do you know, is it... It's Clang11.

Is it... Yeah, is it part of Clang? Oh. Clang11. is it? It's Clang. It's Clang 11. Is it, yeah, is it part of Clang? Oh. Clang 11. Who knew? There's always some new tool to learn about in the Clang suite of tools here. Yeah. Today's sponsor is the PVS Studio team. The company develops the PVS Studio Static Code Analyzer, designed to detect errors in the code of programs written in C, C++, C Sharp, and Java.

Recently, the team has released a new analyzer version. In addition to working under Windows, the C Sharp part of the analyzer can also operate under Linux and macOS. However, for C++ programmers, it will be much more interesting to find out that now you can experiment with the analyzer in online mode on the godbolt.org website. The project is called Compiler Explorer and lets you easily try various compilers and code analyzers. This is an indispensable tool for studying the capabilities of compilers. Besides that, it's a handy assistant when it comes to demonstration of code examples. You'll find all the links in the description of this episode. Basit, how long has the library been available? And you know, are there any, you know, are there any applications using it in production code?

It's been available for three months now. So I've not been up to date with the community to even know which projects are using it or not. But we have a very small user base for now. I still experiment and try to add more features. Okay. That's pretty good. 278 stars already on GitHub for a project that's only a few months old.

Yeah. So are there any other features of STX that you would like to mention that we have not talked about yet? Okay. About the Docs Engine documentation, I had to write some JavaScript and HTML and CSS just to get a very meaningful documentation that's readable prior to the other messy one.

Okay. Oh, I did notice that you're using Abseil, and I'm curious what features of Abseil you're taking advantage of yeah so for the backtrace library the STX backtrace library is more of a wrapper around Abseil's backtrace library

which is Abseil's backtrace library is very low level very can be quite difficult to deal with. SX is just around Abseil's backtrace library. I didn't even realize Abseil provided

a backtrace library. Yeah, I don't think I knew about that either. For the backtrace libraries, I've tried quite many. I've tried LibOnwind. I've tried LibBackactress and many others. That's interesting. I mean, we just had the author of LibBactress on two episodes ago.

Oh, yeah, me and Taylor. Yeah. It's easy to see on the web, too, which makes it very difficult to do it. Okay. So you give a much more high-level C++ kind of access to a backtrace reporting then. Yeah. I wanted to ask about some of your other open-source work. You have done some contributions to TensorFlow, is that right?

Yeah. Can you tell us a little bit about what you've done there? TensorFlow on projects they were there with this arm cortex and protect processors They have this specialized sendy and address related infrastructure assets I mean to which we are trying to take advantage of or to playing computations For this really really constrained embedded devices. So it was really an exciting opportunity that will bring a lot of value to businesses and people. So my contributions were to implement some of these kernels and computation graphs that were used to run all these tiny devices.

So yeah. Okay. So is that TensorFlow or TensorFlow Lite? I think we talked a lot about tensorflow light when we had Andrew Sully on the show yeah

they are not tensorflow light micro it's called tensorflow light micro tensorflow is divided into tensorflow

tensorflow GPU tensorflow light and tensorflow light micro oh okay a lot of tools going in here so many variations

are there any other libraries you contributed to that you want to talk about Oh, okay. A lot of tools going in here. I know, so many variations. Are there any other libraries you contribute to that you want to talk about? Yeah, I also contribute to CMSIS. The CMSIS I mentioned earlier. Right, can you tell us a little bit more about that, CMSIS? Yeah, so it's also like a backend for TensorFlow, which is the one that actually runs the computation. So these are mathematical operations.

They're also for the SIMD instructional sets. So it's more closely tied to assembly. So we use the SIMD instructional sets to perform the computation on it. So the statistical and analytics function just write them for the platform. Well, I do notice here also on your GitHub that you've got several small projects that are related to hardware.

And this seems to then also tie into what you were just talking about with your contributions to TensorFlow. It seems you have an interest in this low-level, small hardware kind of thing. Yeah. I've been trying to learn more about operating systems for about two years now.

So I basically started from writing drivers, so very low-level drivers. I've been writing them for small, tiny platforms. So you said you're currently majoring in mechanical engineering. How far into your degree are you? I'm in my penultimate year now. Okay. Four or five years. So no chance you're going to switch to computer science at this point, probably.

No, at all it does sound like you do have a really deep interest in hardware operating system c++ that kind of thing yeah okay uh i know you had an internship at goldman sachs over the summer i wanted to ask how that went yeah that was a really really big opportunity for me. I got to learn about the finance sector and how to write software. It's totally different from the normal developer-facing software. Do you have any recommendations for people who are looking for internships? I think the biggest recommendation I have is not to give up because rejection is part of the process. I spent about two years

searching for internships and I can't really find any. And the best people you can find, you can reach out for help is your network. You have to try to meet more people. You never can tell out for help with your network. You never try to meet small people. You never can tell who can help you. And was this an internship that you participated in remotely

or did you go to a Goldman Sachs office during this time? Remotely. Oh, okay. Was it only remote because of the current conditions with the pandemic or would it have been remote either way? Yeah, it's a pandemic. It's normally 10 weeks, but we spent five weeks in a pandemic. So you are definitely the first guest we've had on who currently lives in Africa. Was your location in Nigeria at all a problem

or an extra hurdle in trying to find an internship? That was my biggest hurdle. The problem wasn't really finding an internship, but finding people to sponsor you out of the country

or have you do it remotely. That was the biggest order for me and most people most people don't want to air from nigeria because they have this impression that most nigerians are fraudsters and all those kind of corrupts yeah so so it's really really difficult i'm glad you're able to get one. Yeah. So when you had been accepted for coming to C++ now, had you already gotten whatever visas you would have needed? Yeah, it was terminated. My application was terminated.

You know, Trump, I think he ended in law that they should stop issuing J G1 visas or something. So my visa application was terminated then. Oh, okay. That's unfortunate. Hopefully we can try again next year or something. Yeah. Do you hope to be able to come to some of the C++ conferences next year, or are you going to try to apply to them?

Yeah, I do hope to. I'll try to apply again next year. I'm looking forward to it. Hopefully I'll be there and we can actually meet in person. I also, back to your open source projects, I did notice that you have a Patreon.

Is that right? Did I get that right? Yeah. All right, cool. So, you know, people can contribute to your help, encourage you to keep working on these projects, I assume. Yeah, sure. I would really appreciate that. All right. Very cool. Is there anything else you'd like to go over before we let you go

boss it um none for now okay all right well it was great having you on the show today uh definitely keep up the good work i'll encourage listeners to check out stx and check out your patreon uh to help support your ongoing development definitely thank you make sure we have links to all that yes make sure we have links to all that. Yes, make sure we have links to all that. And where can people find you online? I know you're on Twitter a lot, right? So what's your Twitter handle?

Basit. Very cool. Make sure we have a link to that as well. Yeah. All right. Thanks, Basit. Thanks for coming on. That was great.

Thank you. Thank you. It was nice. Thanks so much for listening in as we chat about C++. We'd love to hear what you think of the podcast. Please let us know if we're discussing the stuff you're interested in, or if you have a suggestion for a topic, we'd love to hear about that

too. You can email all your thoughts to feedback at cppcast.com. We'd also appreciate if you can like CppCast on Facebook and follow CppCast on Twitter. You can also follow me at Rob W. Irving and Jason at Lefticus on Twitter. We'd also like to thank all our patrons who help support the show through Patreon. If you'd like to support us on Patreon, you can do so at patreon.com slash CppCast.

And of course, you can find all that info and the show notes on the podcast website at cppcast.com. Theme music for this episode was provided by podcastthemes.com.