One of the strengths of Aspire is still to be able to make this happen
git clone x
dotnet build/run y
Where y can be a very, very complicated distributed application
1
231
I am not saying it should replace all these tools technically (which is silly and not possible), but more like a general concept. Devs dont understand and want to understand the difference between dotnet run and aspire run for example
1
227
Replying to @davidfowl @mattwatsonkc
I think he has a valid point. Aspire should become a default concept and a ‘commodity’ just like msbuild, csproj, roslyn etc. Even better: Aspire should ‘replace’ all of this as the home for all these tools as the project, build and deployment system for .NET™️
1
1
238
Replying to @GerjanDev @davidfowl
Nr1 is esstential for our company, nr2 will attract many, many more devs, even outside the .NET space. Its a universal problem you are addressing (see Karpathys tweet)
/cc @DamianEdwards
33
Replying to @davidfowl
I think its great and I believe in it. However, to make it a big succes you need two things
1. Better support for legacy apps, .net 4.8 etc
2. Support typescript as well for describing the aspire topology
1
102
Next idea for @davidfowl: create a code first build system in Aspire.
Mark MSBuild as legacy/obsolete. Stop the XML magic, way too complex resolvers and relative slow build times.
1
202
Make it crazy fast, cs based instead is csproj based (dotnet build MyBuildInAspire.cs).
Free lunch: Finally the equivalent of javac, gcc, go build, rustc, swift [srcfile]
Yes, this would be a breaking change and I am totally fine with it. Make .NET cool/modern, with Aspire
2
65
Next idea for @davidfowl: create a code first build system in Aspire.
Mark MSBuild as legacy/obsolete. Stop the XML magic, way too complex resolvers and relative slow build times.
Replying to @karrisaarinen @linear
Love it. Till now your big updates are spot on, which is very impressive and ready well done. Cant wait for the next big one, can you say something about that?
My hope: top level documents, time tracking and public pages (share roadmap and docs)
128
Replying to @davidfowl
1. For our monolith
2. For our local microservices
3. For our global microservices
We have a hybrid cloud saas solution, 15+ year old
If you mean it should he GH Actions => Nuke as correct order you are formally right since GH Actions is the runner that will invoke Nuke
724
Replying to @davidfowl
1. Nuke => GitHub Actions => Octopus Deploy => Private Cloud
2. Dotnet cli => GitHub Actions => Octopus Deploy => Private Cloud
3. Dotnet cli => GitHub Actions => Azure XYZ
3
9
4,613
Replying to @GerjanDev @davidfowl
A local runner that can generate YAML files would be even nicer 😍
20
Replying to @GerjanDev @davidfowl
I used to be an advocate for agnostic (local) ‘build’ scripts but I sort of gave up on that 😢
A better local github workflow runner would be nice
1
40
Replying to @GerjanDev @davidfowl
So now we are stuck in YAML files. Still not like it but like I said the native integration with GH concepts wins (envs, vars, secrets, deployments, hooks, PR checks etc).
Same for gh workflow concepts like parallel jobs, context, caches, multi runs, security etc
1
38
Replying to @GerjanDev @davidfowl
But the amount of layering seems overkill as well to keep Nuke. For example a simple .net app: gh workflow yaml file, gh actions, gh runner, nuke, msbuild, roslyn… and going reverse for errors and logs.
1
49
Replying to @GerjanDev @davidfowl
Yet I still love Nuke: console app, nuget, c#, local, repeatable builds, VS support, debugging etc
But the fact that you almost need to write C# for new/changed projects seems like a horde. For small projects it feels overkill and its another tool to learn (cognitive overhead)
1
64

