Thanks!

i would also like guide/demo to run a project inside docker.

Yep you can run Dev containers anywhere. You can also tunnel into the remote machine then launch your containers as if your on that machine with no setup. Just download the pack of remote dev extensions. Plenty of good docs on them too.

Yep, ROS 2 uses DDS underneath, it's a bit complicated and there are a few different implementations that can be swapped out (referred to as middlewares).

you can always use vim to manage your 300 sourcefiles and depedencies, want to code like in the 1980's with a terminal and jump back and forth to man pages - go for it

@@guruware8612 'jump back and forth to man pages'? You don't know how to use vim/neovim yet 😊

You still have to manage your 300 dependencies in VScode with JSON instead of lua

@@guruware8612 Never had a proper vim config eh?

​@@guruware8612blud thinks terminals are only for the 1980s 💀

Yes it's great for any kind of development where you are regularly changing environments (e.g. need to support different distributions)

They charge for visual studio, so I think they know. Besides the ide is just the gateway drug from their other services, like github copilot.

They are already indirectly with github copilot

Shhh 🤐

Just use the open source one

@@RajinderYadav VSCode is not even a code editor, it's just a glorified text editor running in Electron which makes it less efficient than any native text editor.

I don't think it's possible at this moment. You can ask Chat GPT/Google bard to do things like this: In C programming language, write me a function that takes 3 arguments: an integer a, a boolean b and a string c then returns another string. Then you describe what your function logic is. If other people can understand what your function logic does when reading what you write, then those AI's might and be able to generate some code for you.

Unfortunately at this moment in time Chat GPT is very prone to just straight up lying about certain aspects of programming implementation - in particular it likes to make up functions that feel like they should be part of external libraries, but actually aren't. You have to already know about the thing you're asking it in order to make sure it's right... which kind of defeats the whole purpose. There are uses for it - but they're not super obvious or straightforward right now.

For any application that has to touch "real" things like databases, hardware or the such, I would heavily advise against trusting ChatGPT with anything.

The trick is that you don't use chatgpt and actually learn robotics code so you know what your building and what it's doing instead of blindly copy pasting from a glorified next syllable predictor

Just never ever use words "effectifely" and "ChatGPT" in one sentence, please. If you can't code - learn it, if you can get a grade on mechanical engineering, it's strange to see you whining about some code which even some kids can do.

Well for starters, we just added a ton of complexity to an already complex subject. When things break, now you are troubleshooting both docker and the underlying OS. Really though...for the average developer, you aren't adding anything except complexity.

It is not really lagging behind. There is/used to be X11 Forwarding, but ironically we are moving away from X11 at the moment.

@@DanielVoogsgerd how is it not lagging behind mate? X11 Forwarding is really buggy and unstable... would you say its on par with RDP? RDP works flawlessly. Probably something Wayland-based will have a decent RDP competitor in the future.

@@rubiskelter I thought you meant that there wasn't really anything. But fair enough, at least nowadays X11 Forwarding isn't really up to par with RDP I suppose. Although I think X11 used to be a lot better in the past, but applications are constructed differently nowadays. Personally I don't really use nor wanna use either 🤷‍♂️

Thanks!

Yes, but now imagine you are responsible for developing tools that have to work across a variety of robotics platforms, each with different environments and dependencies and quirks, and now you don't need to have a whole VM for each one. (Which is the current practical alternative in industry). I'd prefer a couple GB docker image to a hard drive full of VM's but each to their own!

@@ArticulatedRobotics I probably didn't get my point across correctly. I neither blame you nor Docker, but rather the industry which brought us the mess with all those dependencies, quirks and VMs. As robotics often involves a second, incompatible hardware platform, that is actually acceptable - but nowadays people use Docker for tiny tools like Pihole and think nothing of it. We got used to ridiculous waste of resources to compensate chaotic development environments. Ironically, one of the few bright spots is the language in which Docker was written in: With Go, you can write self-contained, tiny binaries and even cross compiling isn't that hard.

I guess there's two answers here. As to why Docker is useful in general, I've made a few videos on the topic. I think it's most helpful for package maintainers or people who are otherwise needing to work on multiple ROS versions or different sets of dependencies from a single machine. For example my main development machine is still running 20.04 and I chop and change between foxy and humble. This video in particular is about why Dev Containers make developing in a container (as opposed to deploying in a container) much easier, especially for people who are newer to Linux/working from Windows.

@@ArticulatedRobotics Thank you for responding.

Yes, I do agree and it hurts me a little every time I have to do it! When I have the time I need to sit down and do some proper A-B testing with more straightforward titles/thumbs but it appears to me that this is just what you have to do to actually get the content viewed at all. My hope is that for most people they are satisfied enough with the result that they'll put up with it - I know that's how I feel about other creators. I try to put something in the description to help with searchability, I wish you could release with two sets of titles, one to drive stray clicks from general interest and one that is more informative for those searching a particular topic. It's definitely something I have to grow in, all I can say is that this video has outperformed my next best (in its first week) by over 4x, and if I called it "VS Code Dev Containers Overview", that wouldn't have happened, but I also don't want to waste people's time so perhaps there's a middle ground. :) Thanks for the feedback!

​@@ArticulatedRobotics fair enough, thanks for replying that's much appreciated. While I am not much on you tube anymore I still enjoy good content from creators like you. my apologies, the comment was a bit too strong, sometime it can get frustrating..

​@@pepposoftwaredev3950 No man, I agree with both of you but your comment was not even the tiniest bit too strong, google (or other search engines for that matter) and youtube search are really near useless for me or atleast like 20 times worse than they've been a decade ago and since they're like the gateway to internet content, the whole internet is suffering heavily from this. And it's not really the creators fault, it's the fault of silicon valley. You can literally memorize and search for the exact title of a (unpopular) video and it still won't show up, which is why I almost bookmark everything I consider revisitable cause I ain't gonna find it that easily by searching anymore.

Yeah it drives me nuts haha but at least it's smaller and quicker than a VM...

Sorry…who “created the problem” of different operating systems existing…? As a DevOps engineer, this is THE way to build things FAST. No offense, but you genuinely have no idea what you’re talking about.

Yeah. What a problem. Being able to compile your code against different environments from one machine. What a problem that is. Dumbest comment I’ve read this week.

@@MiningForPies It’s not just about compiling code for different environments, it’s about the environments themselves. You neglect those environments, you will fail, within each for a different reason, and you will not know why, and you will be overwhelmed. I take it you’ve never had to install proof-of-concept software on-prem for differing environments. Clock yourself the next time you set up a PostgresQL server. Guarantee I’ll do it 8x faster at minimum.

@@astrovicis why are you saying that to me. I’m for this method.

@@astrovicis and I’ve been installing proof of concept software on premises for 25 years. Including one client who has a none internet connected set of servers with 900GB worth of databases, who are limited to secure USB data transfer (max size 6GB). Don’t you fucking come at me with having no experience.