New places (no content yet): • https://t.co/5jGSUbhrnv •

This works for method visits as well, e.g. here is a method that gets called two times. You can see what called it that time, and what return value was produced for that specific call.

Recently I was thinking about "time-travel" alternatives to traditional debugging for SharpLab's continuous Run. Here's one outcome. SharpLab used to dump all loop values at once, but that wasn't too useful. Now you can select a specific iteration and see values and flow within.

Happy to see !! feedback considered and !! excluded for now. As someone said in previous comments, !! would make "string x" obsolete — we would always use either "string? x" or "string x!!". Personally I am still keen for <Nullable>enforce</Nulable> but I do get the concerns.

As a Russian, I want to say: please listen to Ukrainians and try to help if you can. This pointless, idiotic invasion war is a (yet one more) black mark on Russia's history.

"Giving up" is an underrated developer skill. 16 hours into a nice refactoring, branch has 300 files changed, 100 will need new tests? Give up. Couldn't convince team your code style is best? Give up. Love personally reviewing every single PR, but the team is growing? Give up.

Just FYI, as of now SharpLab has been migrated to .NET 6 (except some older branches).

Going to enable a very aggressive (and very overengineered) cache on SharpLab, please report any issues.

Regular, 1x dev: Decides to add a cache. Adds a cache. Me, 0.01x dev: Persistent cache? What if someone puts PII into SharpLab, must encrypt at rest. Browser is untrusted, must write from server. Can ask branch API, what if URL is faked? Days later, lots of learnings, no cache.

Just to add -- this is a concept for commands at solution level, at project level you probably want to run the project itself.

And yep I know I can do a custom tool and call it e.g. run2, but I liked the symmetry between npm run and dotnet run here, and it was an interesting challenge.

I was curious if you can achieve something like npm scripts (arbitrary commands) with dotnet run -- and you can! Only thing needed is this small csproj in top solution folder: https://t.co/oLLVtvAjak (This is a proof-of-concept, not fully cross-platform, not polished at all)

I did consider it, I wish I could actually do it though😂

* Some limitations apply

Tired of using SharpLab to decompile C# to IL? No problem! Thanks to great work by @konradkokosa and @vivekmapara you can now decompile IL to to C# instead. https://t.co/uQOpyvY43S

A tech pattern to avoid: automatically using latest version. jQuery CDN's -latest got fixed at v1.11, PowerShell's StrictMode -Latest might break your scripts in the future, Azure/arm-deploy GitHub Action picks GH VM's /latest az cli — which just broke it for everyone.

Writing low-allocation code in new .NET (Memory/Spans/Pools) is quite fun. (Side-projects only, too complex for work) It builds a mindset: looked to replace ArrayPool with MemoryPool, but MP Rent() returns a disposable ref type — suspicious. Checked code, yep, always allocates.

Thanks to everyone who voted for this!💎 I very much appreciate both the outcome and the message that SharpLab is useful to all of you.

6. But overall generators are quite fun to use, and I love to see this progress -- on the other hand I did feel that something like mapped types (TypeScript) could have solved a lot of my work without source generation.

5. Using attributes to setup generators is a bit awkward. I would love to have Get<I> giving you MockOfI, or Query("select id, name from client") giving you (id, name) without any extra setup. But that requires C# to overload by specific generic type or specific string literal.