CppCon 2016: Dan Saks “extern c: Talking to C Programmers about C++”
Video Statistics and Information
No captions available for this video.
Channel: CppCon
Views: 130,564
Rating: 4.6636004 out of 5
Keywords: CppCon 2016, Computer Science (Field), + C (Programming Language), Programming Language (Software Genre), Application Software (Software Genre), Bash Films, @BashFilms, Conference Video Recording, Event Video Recording, Video Conferencing, Video Services, Dan Saks
Id: D7Sd8A6_fYU
Channel Id: undefined
Length: 96min 26sec (5786 seconds)
Published: Sat Sep 24 2016
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.
I didn't think I'd be interested, but I ended up watching it all the way through-- even to the end of the comments!
One thing this talk ought to have addressed was the litany of C++ features that must be disabled to be practical in an embedded microcontroller environment. Furthermore from the way he was talking, I am assuming he was referring strictly to MCU programming, although these days the term embedded can refer to devices with DRAM and an MMU and/or full OS (where C++ with all the bells and whistles is already use very widely). For an MCU environment where you have 2-32 kB of ROM, there are many features that must be disabled to even have a snowball's chance in hell of fitting in the nonvolatile program memory.
He touched on many of these: virtual functions, run-time type information, multiple inheritance, but there are other considerations, too, like the fact you often can't use dynamic memory allocation, or would have to manually overload new and delete if you do. If you approach "talking to a C programmer" like a door-to-door missionary in the first place, people like myself who use both languages have to be honest about/"be able grok" the limitations MCU programmers are going to be dealing with, and not condescendingly propose solutions from the desktop/server that really make little to no sense on the platform. I think the subset of advantages he was focusing on were good. In particular template metaprogramming can ease the macro hell MCU programming can quickly become and generally make platform-specific customization and compile-time checks more programmatic. If he wants to encourage people to proselytize, though, it is necessary that C++ programmers realize the (still useful) but somewhat limited subset of features microcontroller programmers will be dealing with.
I have just started K&R to broaden my access to embedded. What do you all think of understanding C to journey into embedded given Dan's points?