Hacktoberfest 2026: những issue maintainer đã đánh dấu cho tháng Mười, đang mở và phù hợp người mới. Xem issue Hacktoberfest

Java → C# translation: an idiomatic C# printer driven by the modification and nullability analyses

Đang mở
  • Có gì không ổn?
#115 0 bình luận 0 reaction 0 người được giao Xem trên GitHub

Maintainer thường phản hồi trong vòng 1 ngày

Chưa có ai nhận issue này.

Đánh giá

Độ khó
5/5
Thời gian dự kiến
Hơn một tuần
Mức phù hợp với người mới
8/100
Loại issue
Tính năng
Độ rõ ràng
Khá rõ ràng
Mức độ hoạt động
Sôi nổi
Công nghệ
csharp, java
Lĩnh vực
compilers

Hướng nghiên cứu

Start by reading the Kotlin printer, maddi-cst-print-kotlin, and its JavaToKotlinRatchet, KotlinPrintMessage and NullabilityVerdicts seam, since the issue says the C# printer mirrors them. The first step is the new maddi-cst-print-csharp module with its TypePrinter/MethodPrinter/FieldPrinter implementations and a csharp-printing.md design doc. The issue does not define a finished state beyond the six-step plan, so a newcomer should not start here.

Do mô hình lập chỉ mục viết ra từ nội dung của issue.

Mô tả

enhancement extension front-end:csharp

Goal

Translate Java to idiomatic C#, automatically: parse with the openjdk front end, print the CST as C# with a new maddi-cst-print-csharp module, and judge the result with the C# compiler. This is the C# counterpart of the Java → Kotlin translation (maddi-cst-print-kotlin, JavaToKotlinRatchet).

This is the printer direction. #111 covers the opposite direction (a Roslyn front end reading C#) and lists a C# printer as out of scope. Nothing here depends on #111's plan: the struct spike, the Predefined refactor and the exporter are front-end work. A Java-parsed CST is JVM-shaped, and the printer starts from that.

What drives it: the analyses

What sets this apart from a syntax converter is that maddi's analyses decide the output:

  • Nullability (NullabilityPass in maddi-mod) decides where ? goes on reference types under #nullable enable, and where !, ?. or ?? appear. The NullabilityVerdicts seam that maddi-cst-print-kotlin uses carries over directly.
  • Modification and immutability decide whether something can be readonly, init-only or IReadOnlyList<T>/IReadOnlyDictionary<K,V>, whether a class can become a record, and whether a value type (readonly struct) is safe. Where the analysis proves code doesn't modify something, the C# says so.
  • Prepwork (getSetField) collapses getters and setters into C# properties, as it does for Kotlin.

Without verdicts the printer still produces valid C#, just less precise: every reference is nullable-oblivious and every collection mutable.

Automation

The translation has to run without manual editing. It is measured the way J2K is: a ratchet per corpus (fernflower first) that translates the main sources, compiles them with dotnet build / Roslyn, and records syntax errors, semantic errors, compiling files and eventually passing tests. A metric that gets worse fails, and one that gets better must be written into the ratchet file. A shape the printer can't translate gets a named message (as KotlinPrintMessage does), never silently wrong output.

Harder than Kotlin

  • No JVM interop. Kotlin can call the JDK and the corpus's original classes; C# can't. JDK references need a BCL mapping or a small runtime shim (java.util.*, String members, streams → LINQ, exceptions, boxing). This mapping is the bulk of the work. It also means there is no "compile against the original classes" stage: the whole corpus has to translate first.
  • Semantic gaps:
    • checked exceptions are dropped;
    • wildcards (? extends/? super) on reified generics;
    • non-static inner classes need an explicit outer reference;
    • anonymous classes become nested classes, or lambdas/delegates for functional interfaces;
    • enums with fields or bodies become classes;
    • Java methods are virtual by default, so C# needs virtual/override/sealed;
    • >>>, integer and char conversions, string switch, labelled break/continue (→ goto).

Plan

  1. maddi-cst-print-csharp: printers on the cst-api TypePrinter/MethodPrinter/FieldPrinter interfaces, reusing OutputElement and Formatter2; a CommonJavaToCSharp test base; a csharp-printing.md design document.
  2. Syntax coverage of every statement and expression form, plus a JavaToCSharpRatchet on fernflower counting syntax errors.
  3. Structure and the JDK → BCL mapping: semantic errors, then compiling files.
  4. Nullability verdicts → #nullable enable output, with a ratchet in maddi-mod (like TestJavaToKotlinFernflowerNullability).
  5. Modification/immutability verdicts → readonly, read-only collection interfaces, records.
  6. The corpus's tests, translated, passing.
Ngôn ngữ chính
Java
Star
1
Fork
1
Merge trung bình
1 giờ 34 phút
Pull request đã merge (30 ngày)
7

Chuẩn bị môi trường

Bắt đầu từ đâu

  1. Đọc hết issue, rồi đọc hướng dẫn đóng góp của dự án.
  2. Bình luận trên issue rằng bạn sẽ nhận — tránh hai người làm cùng một việc.
  3. Fork repository và làm thay đổi trên một nhánh.
  4. Mở pull request có tham chiếu số hiệu của issue.

Issue khác của CodeLaser/maddi

Tất cả issue của CodeLaser/maddi

Issue tương tự

Thêm issue về Java

Nhận issue mới trong hộp thư của bạn

Bản tóm tắt ngắn những issue GitHub phù hợp với người mới.