FoBiS — Fortran Building System
Automatic dependency-resolving build tool for modern Fortran projects — no makefiles, no boilerplate. Package manger highly integrated with GitHub, install and fetch project dependencies. Add instrospective doctests to your Fortran project and exploit AI-powered programmatic buildings with the provided AI skill. Manage a whole ecosystem of interconnected repositories: dashboard, cross-repo checks, release train.
| ⚡ Zero-configuration builds Drop FoBiS into any Fortran project and run fobis build. It scans sources, resolves all use, include, and module dependencies automatically, and compiles in the correct order — no makefiles, no boilerplate. Convention-based discovery finds src/, source/, and app/ directories automatically. First tutorial |
📄 fobos — the FoBiS makefile A concise INI-style configuration file replaces makefiles entirely. Define multiple build modes, templates, variables, custom rules, feature flags, multi-target sections, and hooks — all in one readable file. fobos reference |
| 🌐 GitHub integration Declare dependencies in a [dependencies] fobos section — fobis fetch clones, pins to branch/tag/rev/semver, and pre-builds them; fobis build picks them up automatically. The commits fetched are recorded in a lock file next to the dependencies. Install any GitHub-hosted FoBiS project directly with fobis install user/repo. Fetch deps · Lock file |
🏗️ Named build profiles & feature flags One-word compiler flag presets ( debug, release, asan, coverage) for every supported compiler. Feature flags let you activate named compile-time options without separate build modes — mix and match freely. Build profiles · Feature flags |
🧪 Integrated test runner & coveragefobis test discovers, compiles, and runs all Fortran test programs — no external harness needed. fobis coverage generates HTML/XML reports via gcovr or lcov and can fail CI when coverage drops below a threshold. Test runner · Coverage |
🔬 Introspective doctests Embed micro-unit-tests directly inside Fortran comment docstrings. FoBiS generates, compiles, and runs volatile test programs automatically — inspired by Python's doctest module, no test harness needed. Doctests |
| ⚡ Build cache Content-addressed cache keyed on source commit, compiler, and flags — reuses object files across clean builds and branch switches. fobis cache list/clean for housekeeping. Build cache |
🤖 Introspect & Claude Code skillfobis introspect --all emits machine-readable JSON of the full project graph — sources, compiler flags, dependencies, targets. Powers the bundled /fobis Claude Code skill for AI-assisted builds right in your editor. introspect · Claude skill |
| 📋 Scaffold — boilerplate sync Keep CI workflows, docs configs, license files, and scripts identical across all your Fortran repos. fobis scaffold status shows drift; fobis scaffold sync fixes it; fobis scaffold init bootstraps new projects. All templates bundled inside FoBiS — no extra dependencies. Scaffold reference |
🧠 LLM commit messages Generate well-formed Conventional Commits messages for staged changes via a local LLM. Supports Ollama and any OpenAI-compatible endpoint (LM Studio, vLLM, llama.cpp) — no cloud account, no API key, no data leaving your machine. LLM commit guide |
| 🧭 Ecosystem dashboard Register your interconnected repositories once; fobis ecosystem dashboard shows them all at a glance — releases due (bump suggested from Conventional Commits), CI at HEAD, scaffold drift, dependency pins behind the upstream heads — in the terminal or as a themed HTML page with a live dependency diagram. Ecosystem guide |
🚂 Check & release trainfobis ecosystem check LIB builds and tests every project downstream against your local, unreleased working copy. fobis ecosystem release releases everything due, dependencies first, refusing to start while any project is blocked and stopping safely at the first failure. ecosystem reference |
🔎 Code qualityfobis lint checks the sources with rules for bug-prone, obsolescent and non-portable code and for OpenMP, OpenACC and MPI defects; fobis review reports only on the lines a change touches; fobis query answers who calls, writes or offloads what; fobis modernize applies fixes only when builds, tests and cases are unchanged. Code quality track |
📐 Verified numerics Declare cases and the quantities they compute; fobis verify compares them with accepted references within declared tolerances, fobis matrix across compilers and checked builds, fobis assess estimates the digits you can trust, fobis diff-run and fobis bench compare two revisions, fobis gate runs them all, locally or under Slurm. Verified numerics track |
| 🧠 AI assistance, grounded Local open models managed by FoBiS explain build failures and crashes, answer questions about the code with citations, generate tests that must build and pass, review changes and propose merge resolutions checked by the compiler; fobis mcp serves the same facts to AI agents. Facts first, model last. The AI design |
|
Why FoBiS?
Modern Fortran's module system is powerful — but tracking inter-module compilation order by hand in a makefile quickly becomes a nightmare as project size grows. Every time you add a module, rename a file, or restructure directories, the makefile needs manual updates.
FoBiS solves this completely. It parses source files on every invocation, rebuilds the dependency graph from scratch, and compiles in the correct order — with no configuration required for simple projects.
# That's it. FoBiS finds all programs, resolves all dependencies, compiles.
fobis build
Starting from nothing, fobis new writes a project that builds and tests at once:
fobis new demo && cd demo && fobis build && fobis test
For complex projects, a single fobos file in the project root replaces makefiles entirely:
[modes]
modes = debug release
[debug]
compiler = gnu
cflags = -c -O0 -g -Wall
build_dir = ./build/debug/
[release]
compiler = gnu
cflags = -c -O3
build_dir = ./build/release/
fobis build --mode release
| zero-configuration build | simplify complex dependency |
|---|---|
| GH integration, project install | GH integration, dependancies fetch |
| ecosystem: graph and dashboard | ecosystem: check before releasing |
Ecosystem — manage interconnected projects
Several Fortran libraries, each in its own repository, depending on each other? Register them once and fobis ecosystem handles them as a whole — read-only unless you ask it to act, and nothing is ever pushed without your confirmation.
fobis ecosystem discover # FoBiS repos under your root not registered yet
fobis ecosystem add PENF FACE BeFoR64 # register them (your config keeps its comments)
fobis ecosystem graph # release order and dependency tree
fobis ecosystem dashboard # releases due, CI, drift, stale pins (--format html for a page)
fobis ecosystem exec -- 'fobis clean && fobis build' # any command, every project, dependencies first
fobis ecosystem check PENF # does my local PENF break anything downstream?
fobis ecosystem release --dry-run # what needs a release, with which version
fobis ecosystem release # the release train
| pre-release impact check | release train |
|---|---|
See the ecosystem guide for the concepts, a one-minute demo you can run locally, and recipes.
Scaffold — boilerplate management
fobis scaffold keeps CI workflows, docs configs, license files, and other boilerplate in sync across all your Fortran repos. All templates are bundled inside FoBiS — no external dependencies.
fobis scaffold status # drift report: OK / OUTDATED / MISSING per file
fobis scaffold sync --dry-run # preview diffs without writing
fobis scaffold sync --yes # apply all updates silently
fobis scaffold init # bootstrap a new project (creates src/, docs/, .github/, …)
fobis scaffold list # list all managed files and their categories
Project variables ({{NAME}}, {{AUTHORS}}, {{REPOSITORY}}, …) are resolved from the fobos [project] section, the git remote URL, and git config. See the scaffold reference for full details.
AI assistance, on your machine
FoBiS runs open models itself — llama.cpp's server, a pinned release, and a model pulled once — so there is no service to install, no account, and no code leaving the machine. Every answer is built on the project's facts (the compiler's message, the code, the interfaces) and cites the sources it comes from.
fobis llm runtime install && fobis llm pull coder-7b # once
fobis build --explain # why the build failed: facts first, then a cited cause and fix
fobis ask -p src "What does simpson require of the number of intervals?"
fobis commit # a Conventional Commits message for the staged changes
Recorded with coder-7b on 6 CPU cores and no GPU: each answer takes about 45 s there (cut from the recording), a few seconds on a GPU. Ollama and any OpenAI-compatible server work too. Guides: explaining failures, asking questions, commit messages, the local runtime.
Showcases
Projects using FoBiS:
- ADAM — Accelerated fluid Dynamics on Adaptive Mesh refinement grids | GitHub
- BeFoR64 — Base64 encoding/decoding library for Fortran | GitHub
- FACE — Fortran ANSI Colors and Escape sequences | GitHub
- FiNeR — Fortran INI ParseR and generator | GitHub
- FLAP — Fortran command Line Arguments Parser | GitHub
- FUNDAL — Fortran UNified Device Acceleration Library | GitHub
- FOSSIL — FOrtran Stereo (si) Litography parser | GitHub
- MORTIF — MORTon Indexer (Z-order) Fortran environment | GitHub
- MOTIOn — Modular (HPC) Optimized Toolkit (for) IO (in fortra)n | GitHub
- PENF — Portability Environment for Fortran | GitHub
- StringiFor — Strings Fortran Manipulator with steroids | GitHub
- VecFor — Vector algebra class for Fortran | GitHub
- VTKFortran — pure Fortran VTK (XML) API | GitHub
Author
Stefano Zaghi — stefano.zaghi@gmail.com · GitHub
Copyrights
FoBiS is an open source project distributed under the GPL v3 license. Anyone interested in using, developing, or contributing to FoBiS is welcome — see the contributing guidelines.
Metadata
Release files for FoBiS.py 3.9.11
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| fobis_py-3.9.11.tar.gz | 1.0 MB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| fobis_py-3.9.11-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 1.9 MB
Release files / fobis_py-3.9.11.tar.gz
| Download URL | fobis_py-3.9.11.tar.gz |
|---|---|
| Size | 1.0 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
b31933a0e14daa78254181d49fa44837e1cc80480e60664f5952082a923804e1
|
|
BLAKE2b-256 checksum How to use checksums |
54a3570b6ba83f7bd6b1b138fcb371a4d2eddf5b0b8428e28c7a6b3826bd1bd7
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.
Transparency logRelease files / fobis_py-3.9.11-py3-none-any.whl
| Download URL | fobis_py-3.9.11-py3-none-any.whl |
|---|---|
| Size | 905.8 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
bf42e4ac97eb99d1a56df57347d2751851e81c597b8245b55bdbec7e1446f5e7
|
|
BLAKE2b-256 checksum How to use checksums |
de8a83f41af97ddd3d6b4ae46495ef67c1e8745eec1c073469d2fe4eb35ce529
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Oct 16, 2026.
Transparency log