Johan Euphrosine
Johan Euphrosine
### What's hard to do? (limit 100 words) Due to the lack of interactive DSLX debugger when troubleshooting a logic issue in a `fn` developers can't really introspect value of...
**Describe the bug** Creating a chan array from an existing set of chans seems to crash `ir_converter_main` ``` [throw_delegate.cc:112] RAW: absl::container_internal::raw_hash_map::at absl::raw_log_internal::RawLog() absl::base_internal::ThrowStdOutOfRange() xls::dslx::ProcConfigIrConverter::Finalize() xls::dslx::(anonymous namespace)::ConvertCallGraph() xls::dslx::ConvertOneFunctionIntoPackageInternal() xls::dslx::ConvertOneFunctionIntoPackage() xls::dslx::ConvertFilesToPackage() main...
### What's hard to do? (limit 100 words) Currently when choosing between the DSLX interpreter, the IR interpreter, the JIT or a Verilog simulator to run a set of DSLX...
### What's hard to do? (limit 100 words) builtins are not organically discoverable by developer: - they are under a (empty) root namespace: code completion will only list them if...
**Describe the bug** By default `interpreter_main` set `--compare` to `jit` which end up compare every test result of the dslx interpreter w/ the jit'ed version of the same function. While...
### What's hard to do? (limit 100 words) The way we set channel fifo depth doesn't feel coherent w/ the rest of DSLX syntax Depth is currently set as the...
Now that most (if not all?) parts of #950 have landed, we should properly document how to use it w/ relevant examples. https://google.github.io/xls/codegen_options/#codegen-mapping already shows the `--multi_proc` option but as...
It doesn't look we currently document how to set channels depth. Ideally it should be mentioned in the following pages: - https://github.com/google/xls/blob/main/docs_src/dslx_reference.md - https://github.com/google/xls/blob/main/docs_src/tutorials/what_is_a_proc.md - https://github.com/google/xls/blob/main/docs_src/tutorials/how_to_use_procs.md And featured in some...
### What's hard to do? (limit 100 words) Currently one can't cast one DSLX array to another one of compatible size (same element_size*element_depth). ``` 0243: fn array_reshape_test() { 0244: assert_eq(...
**Describe the bug** It seems that range expression don't type-coerce into properly annotated typed array but instead redefine the type binding to be the max size of the range. **To...