Welcome to the first blog post of my GSoC 2025 journey! This summer, I’m working with the Fortran-lang organization to compile Fortran Package Manager (FPM) with the LFortran compiler. The project aims to enable full FPM builds using LFortran by resolving parsing, semantic, and compilation issues. FPM has some code dependencies which needs to be compiled with LFortran. The first step is to compile the toml-f library, which is a TOML parser written in Fortran. This week, I focused to compile toml-f with LFortran.

Work done in this week:

• Created workarounds in the toml-f codebase to enable partial compilation with LFortran. Creating workarounds in the codebase is a temporary solution to allow the code to compile and run with LFortran, even if it doesn't fully support all features of the language. Eventually, we will remove these workarounds and make the codebase fully compatible with LFortran.

  I have created a local branch with workarounds: Branch with Workarounds

  Some of the workarounds include:

  1. Here BinOp inside a string reference covered with a ArrayConstructor was causing error. But doing this workaround, we can compile the code with LFortran. However, we need to fix this so I have created a minimal reproducible example (MRE) for this issue.

     Workaround: LFortran MRE: Refer Issue 7277

  2. Here passing Array BinOp (of types A + B(:), where A is scalar and B is array) as a function argument generates an error. I did below workaround and created MRE for this and move forward.

     Workaround: LFortran MRE: Refer Issue 7297

  3. Here issue is ordering of len and kind parameter inside character variable declaration. In LFortran by default length is second argument and kind as first, but it is opposite in real. So I have worked it around and created MRE.

     Workaround: LFortran MRE: Refer Issue 7278

• Creating MREs is crucial step as it provides a simplified version of the code that reproduces the issue. This allows developers to focus on the specific problem without the complexity of the entire codebase. I have created MREs for the workarounds and reported them to LFortran upstream.

  MREs created: MREs for toml-f

• I have created around 10 workarounds and around 15 MREs this week. For both I have given links above.

I would like to sincerely thank Ondřej Čertík and Pranav Goswami for their continuous guidance and support. Also I would like to thank LFortran team members who fixed some MREs I posted which makes a speedy progress.

Plan for next week:

• As many blockers in compilation of toml-f is due to weak areas for Classes and Structs in LFortran, so many MREs are related to classes and I will try to fix them as these are the issues which are hard to workaround. Some Issues regarding classes: Issue 7249 , Issue 7250 , Issue 7253
• Also I will try to generate more MREs from toml-f compilation.