Ondřej Čertík

It seems the following two declarations: ``` character(n) :: x(M) character :: y(n, M) ``` are equivalent in memory: both are contiguous blocks of memory of size n*M, and the...

The String types will mirror the arrays, in the previous comment sense https://github.com/Pranavchiku/lfortran/blob/e34d2c53a340e9c6905950cc8a76b511481cf4c8/doc/src/asr/asr_nodes/expression_nodes/ArrayPhysicalCast.md. FixedSizeArray -> FixedSizeString PointerToDataArray -> PointerString UnboundedPointerToDataArray -> CString DescriptorArray -> DescriptorString Ownership (the same for all...

Note, we can do: ```diff --- a/src/libasr/ASR.asdl +++ b/src/libasr/ASR.asdl @@ -197,7 +197,7 @@ ttype | UnsignedInteger(int kind) | Real(int kind) | Complex(int kind) - | String(int kind, int len, expr?...

Current design: * Use just two physical types: DescriptorString and CChar * At runtime in LLVM we are storing the DescriptorString as an LLVM struct `{*i8, i64}` (pointer and length,...

TODO: * [x] Finish the string test cases, send to main, merge (https://github.com/lfortran/lfortran/pull/7877, https://github.com/lfortran/lfortran/pull/7889) * [ ] Update https://github.com/lfortran/lfortran/issues/7006#issuecomment-2867013984 (or create a new document) to describe every BindC case what...

Mini TODO: * [ ] Submit test cases that only work with gfortran (for now) into main as a PR. Add every combination of allocatable/non-allocatable strings, longer argument than parameter,...

Latest iteration of the design: https://github.com/lfortran/lfortran/issues/4263#issuecomment-2869944911.

Here is a summary of our call: * Add a compiler option `--new-classes`, and if enabled, start implementing this new design in the backend. * Add some reference LLVM tests,...

Yes, I need to think about it. This is very low priority, I think we'll just remove it from fastGPT. I initially thought I would set more variables this way,...

Duplicate: https://github.com/lfortran/lfortran/issues/1851