Replies: 7 comments 15 replies
-
It could be a project file switch too, if the SDK started emitting global usings along with assembly attributes, or if you use a community NuGet package that does the same thing. |
Beta Was this translation helpful? Give feedback.
-
|
It's always been a project switch in VB.NET (via command line args) so there's precedent and an audience of developers that can relay their experiences with it. Personally I liked project-wide imports in VB.NET and didn't find them confusing at all. |
Beta Was this translation helpful? Give feedback.
-
|
you could possibly make it like a project switch by adding a source file via msbuild the same way assembly attributes are added. |
Beta Was this translation helpful? Give feedback.
-
|
(That's what I meant by emitting global usings along with assembly attributes.) |
Beta Was this translation helpful? Give feedback.
-
|
Was replying to HaloFour, I mean at that point it practically make no difference. Even without going through msbuild tricks to insert source, you can just link a file in directory build props. |
Beta Was this translation helpful? Give feedback.
-
Just curious was ever a subset of that considered? That is additions only but extended to everything not just partials. This would enable for example automatic code instrumentation for testing or raising events in more complex scenarios than just INPC. |
Beta Was this translation helpful? Give feedback.
-
|
I think there was a proposal for allowing only one part of a partial type to be marked as partial, with source generators specifically in mind. The net result would be the same as if partial was completely optional. |
Beta Was this translation helpful? Give feedback.
-
|
@HaloFour i believe this does NOT cover everything im talking about. Eg adding a call to diagnostic methods on start of existing method and its end (either right before all returns or before last |
Beta Was this translation helpful? Give feedback.
-
|
It's considered modifications if it involved changing the body of existing code, even if what you were doing was "adding" to the beginning or end of the method. That was really the only scenario that would have been supported by |
Beta Was this translation helpful? Give feedback.
-
|
I completely agree, though. AOP scenarios would be an amazing killer feature for source generators. |
Beta Was this translation helpful? Give feedback.
-
|
Sad to see expression trees stuck in limbo. They're blocking language support for Roslyn syntax trees by their mere existence and are popular enough to be sustained indefinitely. |
Beta Was this translation helpful? Give feedback.
-
|
Forgive my ignorance, but how do they block Roslyn syntax trees? |
Beta Was this translation helpful? Give feedback.
-
|
By existing. If they didn't exist, everything would be using Roslyn syntax trees instead, and there would be demand for better support at the language level. But now everything except source generators already uses expressions and migrating away from them is a daunting task. |
Beta Was this translation helpful? Give feedback.
-
I really don't think that's the case. For one thing, Roslyn syntax trees are C# and VB specific. We wouldn't expect the abstract expression language to be concrete C# syntax trees. Furthermore, our syntax trees are are jsut that... syntax. Expression trees are AST that contain important semantic information in them. |
Beta Was this translation helpful? Give feedback.
-
|
Would a versioned API work for adding features to Expressions? As-in, by default you get the current features (since versioning never existed prior). To get a new feature set, you explicitly opt-in to that version? |
Beta Was this translation helpful? Give feedback.
-
|
@333fred I think this was mentioned briefly in the meeting. |
Beta Was this translation helpful? Give feedback.
-
|
There is potential, but cross-assembly dependencies are a problem. What happens if assembly A generates trees v1, and B consumes v1, then A is updated at runtime now produces v2? |
Beta Was this translation helpful? Give feedback.
-
I wonder if any sort of auto vectorization could be possible. That may be closer to area of theoretical PL though. |
Beta Was this translation helpful? Give feedback.
-
As C# evolving, mixing C# with VB shows less value since VB is almost stopped. Mixing C# and F# aren't easy to use because their programming models are too different. Mixing C# and IL may be useful. I'm interested in exposing things valid in IL but not allowed in C#. Generally, I'm treating mixing C# with VB a proof of concept and architecture, that "we have a compiler model can handle two languages simultaneously". |
Beta Was this translation helpful? Give feedback.
-
The one place where I can see such mixing being valuable is when migrating a larger VB project over to C# iteratively. Given the evolution of VB has effectively stopped that is something that I imagine organizations would be interested in. |
Beta Was this translation helpful? Give feedback.
-
|
and you could always migrate to c#8 under a current runtime, then roll forwards, even if in the future we complicate the migration. Microsoft is happy for such migration paths in many places, going from version A to G requires intermediate steps of C and E. |
Beta Was this translation helpful? Give feedback.
-
A few points i would avoid referring to Fody as "source replacement" since that incorrectly describes what it does, and would lead to have an incorrect understanding
|
Beta Was this translation helpful? Give feedback.
-
https://github.com/dotnet/csharplang/blob/main/meetings/2021/LDM-2021-04-07.md
Agenda
Beta Was this translation helpful? Give feedback.
All reactions