Live data from Hacker News

The Future of the Gitlab Web IDE

about.gitlab.com

111–120 of 168 posts

Re: The Future of the Gitlab Web IDE

#111
post #72

Earlier quoted context omitted.

Maybe a stupid question, but doesn't the remote environment run basically the same software? Or at least the same software as a local container or VM would? If so, why would this break less often?

200 engineers means 200 subtly different laptop environments, each of which could break in slightly different ways. With a good remote environment solution, each engineer can click a link and get a pre-tested, standardized development environment. They can then break it in any way they like... but then they can click that link again to get a fresh, working environment again. Imagine if any time you broke yoir environ…

So why remote instead of a container or VM?

Re: The Future of the Gitlab Web IDE

#112
post #5

While this is cool, for Node.js there's something far better: https://stackblitz.com/ . This is one of the more impressive IDE in browser experiences I've had. Comes complete with a terminal and a shim around `npm` that allows you to install JavaScript dependencies directly to the browser. Some truly cool stuff.

There is also https://www.gitpod.io/ I like the current WebIDE. It has two features which are important to me: - Lauch it with the . hotkey, from the file or merge request you currently browse - multi cursor with CMD+D on a Mac I don’t think I need VSCode, but we’ll see.

Hey, Geoff here from Gitpod. Thanks for letting more folks know about us. If people have any questions let me know.

Re: The Future of the Gitlab Web IDE

#113

Running VSCode remotely introduces many interesting challenges, mostly due to security. (You are letting your users run any arbitrary code on your infrastructure.) However, I guess GitLab already has this sorted out, because of their existing IDE, and they are experts in this area. Also the open source versions of VSCode don't have access to the official VSCode marketplace. (Both VSCodium and coder.com Code Server.)…

It's all dockerized and security hardened probably. I'm guessing this is not that much of an issue. At least not any more than the likes of AWS, Google cloud, etc. allowing third parties to run whatever on their infrastructure. Not to say that there aren't any challenges but they aren't new challenges. And of course Gitlab has long been running CI infrastructure which would have the exact same issues.

Yeah I know, but I have some indirect experience with this (not at gitlab, of course) and it was hard to figure all out.

In short, it's hard to tell if you are doing repeated C++ builds or mining cryptocurrency.

But yeah it's not that different from CI.

Re: The Future of the Gitlab Web IDE

#114
post #5

While this is cool, for Node.js there's something far better: https://stackblitz.com/ . This is one of the more impressive IDE in browser experiences I've had. Comes complete with a terminal and a shim around `npm` that allows you to install JavaScript dependencies directly to the browser. Some truly cool stuff.

Oh hey, a recruiter made me code on stackblitz for an interview recently, and I couldn't find the site again afterwards. Thanks for the link.

Yeah, stackblitz is amazing. At least during that one interview, it worked flawlessly.

Re: The Future of the Gitlab Web IDE

#115
post #97

By the way, Monaco is 10MB+. That's quite impressive nowadays for something so feature filled, but you're staying for 10MB when doing quick edits. Monaco is a mature, powerful and pluggable, which makes it a good platform to build on top of. More cynically, it's a good bang-for-the-buck project if you are seeking a promotion. The alternative is building your own. Maybe that doesn't make good business sense. And that'…

> under 14KB ( a single TCP roundtrip ) That's not a single roundtrip in any TCP stream I've heard of... Still, a lot smaller than 10MB of course.

https://tylercipriani.com/blog/2016/09/25/the-14kb-in-the-tc...

Re: The Future of the Gitlab Web IDE

#116
post #16
post #12

Maybe I'm getting old, but I don't want to do all development-related tasks in my web browser. As a software engineer, I want local control, where an internet connection is optional. I want to be able to experiment with the software in various local setups. I just don't get why Gitlab thinks this is a good idea worth a huge amount of engineering effort. Are people really clamoring for this feature?

You are not alone. If I was made to use this, I'd probably retire early or change careers. Trends towards centralization and uniformity make me sad.

I don't think I'd want to work somewhere where everyone was told what tools to use, 'how to do their job', etc. - even if it was exactly how I happen to like to do it (vim with LSP, but most other IDE-like stuff from the shell directly).

Re: The Future of the Gitlab Web IDE

#117

Earlier quoted context omitted.

It's all dockerized and security hardened probably. I'm guessing this is not that much of an issue. At least not any more than the likes of AWS, Google cloud, etc. allowing third parties to run whatever on their infrastructure. Not to say that there aren't any challenges but they aren't new challenges. And of course Gitlab has long been running CI infrastructure which would have the exact same issues.

Yeah I know, but I have some indirect experience with this (not at gitlab, of course) and it was hard to figure all out. In short, it's hard to tell if you are doing repeated C++ builds or mining cryptocurrency. But yeah it's not that different from CI.

What does it matter if you pay for it?

Re: The Future of the Gitlab Web IDE

#118

I feel that moving from Monaco to VS Code OSS (which is based on Monaco) is analogous to moving from a custom protocol on top of tcp to the standard http protocol. It's a big change, but not necessarily groundbreaking or noteworthy.

Make a free Gitlab account and try out their Web IDE as it is now, and then compare it to Gitpod or just run Coder-server on your machine. It's a humongous change.

The current Web IDE is not an IDE at all - VSCode is.

Re: The Future of the Gitlab Web IDE

#119

It's interesting to me that GitLab is adopting a Microsoft product (VS Code) and Microsoft owns a significant competitor in GitHub. Nothing intelligent to say about that other than to wish I'd been a fly on the wall for the discussions about that. > Next, we asked ourselves the question: Do we want to continue to invest in implementing custom features for the Web IDE that ultimately deliver the same value as those al…

Yeah, especially when they could use something like Theia[0], which I believe was built for this kind of purpose? They both still use Monaco under the hood AFAICT. I'd be interested to know if they considered it and decided not to use it for some reason.

0: https://theia-ide.org/

Re: The Future of the Gitlab Web IDE

#120
post #54

Earlier quoted context omitted.

> Yes. Because local development environments...break ALL THE TIME. This seems like more of a process problem than a technology problem. Our company has a simple rule for this: you break it, you fix it -- immediately. As a result, our development process is very stable.

You break it, you spend days on fixing things instead of being somewhat productive, and if you break the codebase, then we wait while you try to fix it, before we need to step in and do it ourselves. This was sometimes my experience working with and leading teams with devs of varying experience but always full freedom. Nobody actually likes it.

If it takes days to figure out problems with the dev/build environment... uhm... you're having a huge problem, regardless of where files are edited.
Post reply on HN