julia

5c659a4f - debugger: share renderer between gdb/lldb, add jl path command, cap output

debugger: share renderer between gdb/lldb, add jl path command, cap output Factor the value-rendering logic out of julia_gdb.py and julia_lldb.py into a shared, debugger-agnostic julia_debug_core.py; the two scripts are now thin adapters (memory reads and debug-info lookups) plus debugger-specific integration, so the rendering can no longer drift between the two. Add Julia-semantics field/index access, since the C expression evaluator cannot look through jl_value_t*: - a `jl` command in both debuggers: `jl v.inner.name`, `jl v.vec[2]`, `jl v.tup.1` (1-based indexing), plus module globals resolved through the binding table and its partitions (`jl Base.have_fma`); `$jl` continues from the previous result, which also works through structs stored inline in their parent - gdb convenience functions `$jl_typeof(v)` and `$jl_field(v, "name")` that compose inside larger expressions, returning jl_value_t* for boxed fields and native typed values for unboxed primitives Also cap the total size of any rendered summary (MAX_OUTPUT, on top of the existing element/string/depth caps) so printing a huge array or string cannot flood the terminal, and fix non-function singletons to render as e.g. `Base.Irrational{:π}()` instead of a bare type name. Assisted-by: Claude Code (Fable 5)