jetbrains rider plugin?

Jetbrains has finally released their full .Net IDE called "Rider". Though I love NCrunch, I can't stomach using Visual Studio any longer, especially compared to Rider.

Any plans to add Rider support for NCrunch?

Thanks.

Hi,

There is a request for this on uservoice at the moment.

Right now there are no plans to implement this. It's a massive piece of work that would involve rewriting around ~30% of the product in Java.

Remco wrote:Hi,

There is a request for this on uservoice at the moment.

Right now there are no plans to implement this. It's a massive piece of work that would involve rewriting around ~30% of the product in Java.

This is one of the saddest news for me this year.

Just like some other people. The only major reason I stick with Visual Studio is NCrunch.

I understand it requires a huge mount of work to implement this feature. At the same time, tons of developers (desperately) want this feature.

So is it possible to start a crowdfunding project on Kickstarter? With enough funding, maybe it's possible to hire a Java developer work full time to implement this?

I remember sometime ago, people requested RavenDB to run on Linux, so they hired one developer work full time just to make RavenDB to work on Linux.

I think it worth funding a Kickstarter project if it could reduce the time we have to wait by 1 or 2 years.

yanglee wrote:
I think it worth funding a Kickstarter project if it could reduce the time we have to wait by 1 or 2 years.

Sorry, I know this is a disappointment for you.

The problem has less to do with resources and more to do with feasibility. If it were just a case of throwing a year of development into such a feature and there being no further cost involved, this would look much more appealing.

Unfortunately, because Rider is written in Java, the entire UI of NCrunch would need to be reconstructed, along with the layer that connects this UI into the engine. Because of extremely tight performance constraints, the UI for NCrunch is deceptively complex and is built almost entirely on custom controls. All of this would need to be duplicated into a Java based equivalent with an intricate communication system that would touch almost all areas of the engine. Every time a feature is added to NCrunch, it would need to be implemented twice; once in .NET, and once in Java. Every time Jetbrains release a new version of Rider, NCrunch would need to be adapted to simultaneously support the new version alongside all prior versions. The maintenance burden of this feature would be tremendous, it would almost double the cost of keeping the project running.

When you consider that at least 70% of the time spent on NCrunch is spent maintaining integration with the tool stack and software developed by other parties, it's easy to see how a significant increase in maintenance would strangle further development. It's possible to get more capacity (i.e. by increasing team size, changing business structure, etc), but there are other costs attached to this, especially in the areas of quality control and the potential complexity of future features. Integration with Rider has the potential to be highly detrimental to NCrunch.

A feature of this size would need to have a really strong business case behind it, or it would need to be in the strategic path for the development of the product. In regards to the former, I am not presently convinced that the potential size of user base on Rider would be big enough to warrant the costs involved in integration. In regards to the latter, it's never been the goal of NCrunch to integrate with everything under the sun. I founded the project to push boundaries and discover new ways of working. It's very hard to do this when you're tied down in a web of integration points floating in a sea of moving parts.

How deep does the integration with the IDE really need to be for NCrunch to work? I would be perfectly happy with standalone version of Ncrunch that works based on changed files and shows results in its own window. This way you would't need to rebuild the UI in a new language.

Businesswise this could be a nice fit with Microsoft's current strategy of opening up the tool stack.

Another aspect is Jetbrains' view of having another continuous testing tool in their IDE, competing with their own continuous testing feature with DotCover (not yet available in Rider though). You wouldn't want to invest huge amount of money/time just to see JetBrains closing the door for you in order to sell more DotCover licenses.

olli17 wrote:How deep does the integration with the IDE really need to be for NCrunch to work? I would be perfectly happy with standalone version of Ncrunch that works based on changed files and shows results in its own window. This way you would't need to rebuild the UI in a new language.

Businesswise this could be a nice fit with Microsoft's current strategy of opening up the tool stack.

This is an option that I've considered. It would also work for other IDEs. Essentially, it means trading much of the user experience for eliminating the need to reimplement the whole UI. Unfortunately, this half-way approach would still mean integrating and supporting every version of the IDE that is released. NCrunch does still need to synchronise internally with the IDE's in-memory changes and be able to provide inline results. Having the UI hosted by a separate application would also create numerous UI frustrations that I think would disappoint many people. I just don't see such a design meeting a level of quality that I would feel comfortable with shipping, even though I'm sure many people would feel this approach to be better than no support at all.

GreenMoose wrote:Another aspect is Jetbrains' view of having another continuous testing tool in their IDE, competing with their own continuous testing feature with DotCover (not yet available in Rider though). You wouldn't want to invest huge amount of money/time just to see JetBrains closing the door for you in order to sell more DotCover licenses.

This is always a risk when integrating with a product built by another (competing) party, though in this instance I don't think that JetBrains would likely close the door on a plug-in like NCrunch. Integrating NCrunch with Rider would be of value for both products. The same situation currently applies to VS, considering the new Live Unit Testing feature.

It is very sad you hear that there are no plans to integrate NCrunch with Rider. NCrunch is my favorite tool for unit-testing, but since the release of Rider I have completely switched to in for development tool because it's performance outweighed the experience of VS + NCrunch. It means that I will no longer upgrade my NCrunch license. And I am sure that there are many developers that feel the same.

Would it be possible to hook up NCrunch Console Runner Output in Rider? Sure it wouldn't be the NCrunch Experience, but I hope this would give us some of the power of NCrunch quickly?

Would it be feasible to act standalone and support the language server protocol? https://microsoft.github.io/language-server-protocol/
As far as I understand it has gained popularity and tools then can have one integration point but work in multiple editors.

https://plugins.jetbrains.com/plugin/10209-lsp-support

Unfortunately, there's no way to provide an acceptable NCrunch experience in Rider without the following:

1. Being able to extract in-memory code text from the IDE
2. Being able to extract in-memory project details from the IDE (i.e. XML, dependency information, etc)
3. Rendering of coverage markers in the code
4. Correct preservation of the coverage marker positioning when the code changes (frightfully hard to do)
5. Navigation options allowing ability to direct the IDE's UI to different areas of the codebase according to commands by NCrunch (i.e. Go to test, go to exception line, etc)
6. Ability to render stable UI windows in the IDE's process, in a way that won't interfere with the IDE's UI or cause massive frustration by overlapping other windows etc.

Unfortunately, even when cutting the feature-set to its bare minimal, it just isn't possible to deliver the above without deep integration with the IDE. That would mean deep integration and maintained testing with all versions of the IDE and all future versions of the IDE. Which leads us back to the original problems with feasibility and maintainability.

Sorry, I don't see a quick and dirty way to do this that will add value for anyone.

IMHO you should reconsider possibilities. You will greatly increase customer base by
1. Including users from OS like Linux and Mac
2. TDD is language agnostic. I believe that if you build Rider plugin then you will be able to build plugin for all set of JetBrains products. So how about NCrunch for Java, Python or whatever..

I would buy NCrunch for Rider. So you may count for 160$ for that task ))

Is there any update on this? We are 2 yeans further now.

It's the most voted item on your UserVoice, together with a standalone version of NCrunch, so I guess it is what most customers want.
I recently switched to Rider as well, mostly because VS seems to slow down on bigger projects, which I am working on for clients now.
NCrunch is alot better than dotCover in many aspects, so I would prefer using it in Rider.

I was speaking with some of the JetBrains devs a couple of weeks back, and they were hinting to running non-Java extensions in Rider coming very soon.

And that nCrunch might be one of the first plugins like this..? 👀

Might all be wishes from the JetBrains crew, but if you have any teasers or early previews, I'm sure a lot of us would love to try it out!

No nCrunch is my #1 reason to not use Rider full-time, and I'd happily get my team a set of licenses for nCrunch+Rider.

TobbenTM wrote:I was speaking with some of the JetBrains devs a couple of weeks back, and they were hinting to running non-Java extensions in Rider coming very soon.

And that nCrunch might be one of the first plugins like this..? 👀

Might all be wishes from the JetBrains crew, but if you have any teasers or early previews, I'm sure a lot of us would love to try it out!

No nCrunch is my #1 reason to not use Rider full-time, and I'd happily get my team a set of licenses for nCrunch+Rider.

Would be really nice! I'm loving Rider over Visual Studio for several reasons. But for some TDD-intensive projects I keep sticked to Visual Studio because I won't be productive without NCrunch.

There is also the request for a standalone NCrunch which is independent of VS but would open up NCrunch for any integration - e.g. similar to language servers somehow. Maybe enhancing the NCrunch console "a little bit" could be the first step.

Example: https://github.com/PowerShell/PowerShellEditorServices

Just want to mention that, as a long time NCrunch User, but stopped using VS since 2020. I'm still missing NCrunch in Rider, creating a specific Session with some Tests and run them all on save is not the same ... hopefully @TobbenTM is right, and that @Remco does not say anything about that is also a good sign for now :)

Today I missed the Churn feature because of some flaky test. So hopefully, NCrunch 5.x will not disappoint in that way. Fingers crossed!

Dotnet build support. Jetbrains Rider support coming soon!
https://blog.ncrunch.net/post/Dotnet-build-support-Jetbrains-Rider-support-coming-soon!.aspx