Hacker News new | ask | show | jobs
by ripley12 1696 days ago
I'm the author of that issue. I think I can be a little more frank here: this is a horrifically short-sighted decision that has me scared for the future of .NET. I have a tremendous amount of respect for the .NET team, and I'm very certain they're not the ones behind this.

I've been incredibly impressed by the hot reload solution that the .NET team has delivered. It's worked remarkably well in preview across multiple platforms and technologies. I've built my own hooks into the hot reload feature, and I've used it to get a tight feedback loop in many different kinds of applications. It's awesome.

I was so excited to be able to point to that feature and say "yes, .NET has a great developer experience - even outside Visual Studio". I haven't always been able to do that. And now... it's going away as a broadly available feature, so that some bean counter can use it to drive usage of a legacy IDE (I love Visual Studio, but let's not kid ourselves about where the future lies).

Immediate feedback (i.e. a fast inner loop) isn't a niche feature - it's an essential element of any creative activity (see: Bret Victor). They are crippling .NET and it hurts to see it.

4 comments

Completely agree with you, and thanks for opening the issue and raising awareness of this.

I too have been using it successfully in preview (having been waiting a long time for it). To drop it so close to the launch of .NET 6 is crazy to me.

It’s difficult to take a charitable view of this. The messaging around “needing to reprioritise” feels very disingenuous to me.

We were making so much progress…

There's a lesson that you may not want to learn here: the guaranteed disappointment in .NET and Microsoft can be reasonably easily avoided by avoiding them.
As someone who suffered through MS and Oracle (Oracle directly killed one of my companies; MS did a lot of collateral damage) in the 80s and 90s, they have been more than nice for the open source community after Balmer. I hope they rectify this one.
While they have been supporting open source (because, if they didn’t, they’d become irrelevant in no time), there’s nothing that says they aren’t just playing a long embrace, extend, extinguish game here. They may never reach the extinguish part because they can’t find a way to lock users in, but, still, they have interests that are still completely opposite to the welfare of an open source community.
It seems unlikely to me: developers have power these days and their voice, threads like this, appear on radars of management. Even if some pencil pusher 'decided' something, there is no real way out in my opinion. This will probably be reversed and they really have issues because of the .net foundation fall out: as much as I wish to be 25 again, this is not the 90s anymore.
I agree. They can lock in some users, but not that many. Their monopoly power is gone and it seems gone for good.
Visual studio is anything but legacy and I haven't found anything better. Where does the future lie?
IMO - Rider. It's way better with its autocomplete, refactoring, renames, code completion, debugging, 3rd party code navigation, overall responsiveness. I've switched and I feel physically disabled whenever I have to do something in Visual Studio again. The only thing that is lacking in Rider is Azure support(plugin is meh), but I don't really use that anymore, we have CI/CD for that.
Did you use Roslynator and Codemaid extensions? both are free and lighter than Resharper.

Almost always when somebody shits on Visual Studio s/he either uses C++

(and there's significant difference apparently between C++ and C#) or used it with Resharper, or without any extension/addon at all.

Just try Roslynator if you haven't used it yet.

I did use Roslynator. It helps with clunkiness of VS and gives you some refactorings but it felt far from Rider. Roslynator also doesn't fix abysmal code autocompletion. Even after the 'AI autocompletion' that VS introduced like 2-3 years ago, it was always just bad, classes you actually want to use are so far down I found myself typing whole class names by hand anyway.
I try Rider ever so often but what specs are your system; it is basically unusable after I load one of our bigger projects... My m1 16gb just grinds to halt and before I had a beefy thinkpad with 32gb and that stalled completely too. While VS has no issue at all.
Are you sure you’re running a native ARM build of Rider? And how are you running _Visual Studio_ (not VS Code) on an M1-equipped machine?
I have a 16gb Surface Book 2 at work and a personal high specced Dell XPS 9570, both are perfectly fine, even with quite big solutions. Rider has fairly long startup because it's indexing _everything_(this is why code completion is so good) but other than that it is perfectly fluid to work with. With VS(2019) Roslynator worked fine but solution would also take a lot of time to load. I have more issues with Teams than any IDE to be honest.
Have you tried Rider for Apple Silicon? You can grab a preview here https://rider-support.jetbrains.com/hc/en-us/articles/440192...
I've heard JetBrains' IDEs have had performance problems on macOS forever from Mac users (is it because of the Java?). They work great on Windows and Linux in my experience.
VS is great for desktop apps since it has the designer and some other tooling, but for everything else Rider is miles ahead, in my opinion.

Especially when it comes to web. Visual Studio still dies of stroke as soon as someone dares whisper the word "razor" in its vicinity, while Rider has no problem understanding Razor templates and providing full autocompletion.

Not to mention Rider is just straight out snappier. And has integrated terminal without any need for extensions.

Unfortunately, I need SQL Project support before I can ever consider switching to Rider. I could almost move to VSCode for everything but MS refuses to implement nested files. Currently I do prefer writing C# in VS though.
ever tried razor auto indent? the nuget pm ux? ...
I understand the huge disappointment but there’s a bit of hyperbole there. Anyway, lets hope they reconsider w/ .net 7. At the pace they’re going, in a years time.
Not hyperbole.

If .NET is going to stay relevant in the long run, it needs a better reputation for cross-platform development and much better tooling outside of Visual Studio. Newcomers to the language expect it to work well in VS Code, and they are frequently disappointed.

This decision moves .NET backward in areas it already needs to do better in, and to add insult to injury it does so by deleting already-working code.

It's disappointing that they want it to be exclusive to VS (that giant turtle) right now, but it's a new Hot Reload feature. For most existing .NET devs, it's business as usual.

It may set a trend in which VS is pushed as the tool of choice to develop in .NET, which would be a total bummer. I'm a bit skeptical about this since they would be damaging their huge investment in VSCode and cmdline tools. As far as I know, Azure is still their priority. What happened to their leadership to cause such a reversal? Pressure to sell more VS licenses? Maybe, but it still seems doubtful.

I'm giving them the benefit of the doubt until they either explain themselves better, or they release .NET 7 without any goodies to the cmdline devs.

As someone who has put the future of my company on .net core this is not hyperbole; this is quite a worrying development. I did not pick Java because .net core has served me well when I build some prototypes 5 years ago (the 15 years before that we did mostly but I sold that company and all my colleagues were tied in the sale, willingly so I could re evaluate tech).