GSoC 2026 Introduction | Jatin Kumar | Interested in stdlib File System Library

Hi everyone,

I’m Jatin Kumar, a 2nd-year CSE student at MAIT, New Delhi. I’m aiming to contribute to GSoC 2026 specifically for the stdlib File System Library.

Why me?

  • C++ & Systems: Strong background in C++, which I plan to use for the ISO_C_BINDING and C-wrapper logic required for OS-level operations.

  • Ready to go: Local environment (gfortran/fpm) is set up. I’m already studying the stdlib-os branch and the current filesystem specs.

  • Goal: Implement cross-platform directory/file manipulation (POSIX & Windows) to make Fortran more “general-purpose.”

Current Status: I’m looking for a “low-hanging fruit” or an unimplemented function in the filesystem spec to write a proof-of-concept.

Mentors (@Arjen,@milancurcic) — is there a specific gap in the current stdlib_os WIP where I can start contributing code or documentation?

Best, Jatin Kumar ijatinydv (Jatin Kumar) · GitHub

5 Likes

Welcome to the forum, Jatin. As for your question: you can check the repository for open issues on the subject of course and I would do the same myself :innocent:, as I do not know them by heart. I also do not know yet what the GSoC will bring for this year. But let’s see what we can do here.

1 Like

Hi Arjen,

I’ve spent some time digging into the open issues as you suggested and found Issue #1109 (Absolute paths in fortran_stdlib.pc).

I’ve analyzed the root cause in export_pc.cmake. It appears that resolve_pc_libs defaults to absolute paths when it encounters IMPORTED targets like BLAS::BLAS. I’m currently working on a fix that leverages pkg_check_modules to use Requires.private instead, which should significantly improve portability for system packagers (like MSYS2).

I’ll be opening a PR for this shortly to get technical feedback from the maintainers. Thanks for the guidance!

1 Like

Hi Arjen,

Following your advice, I spent some time investigating the open issues and identified Issue #1109 (Absolute paths in fortran_stdlib.pc).

I’ve analyzed the root cause in config/export_pc.cmake and have opened a Pull Request with a fix. My implementation introduces a portable path-to-flag fallback mechanism and leverages pkg_check_modules to ensure implementation dependencies (BLAS/LAPACK) are correctly moved to private fields, significantly improving library relocatability.

Pull Request: build: fix absolute paths for external BLAS/LAPACK in pkg-config export by ijatinydv · Pull Request #1118 · fortran-lang/stdlib · GitHub

I’ve verified the fix on a Windows/MinGW-W64 environment with external OpenBLAS. I’d appreciate any technical feedback you or the other maintainers might have!

1 Like