Software configuration file#

Introduction#

This page constitutes the Software configuration file (SCF) for the eo-data-embedding project, describing the contents of the software configuration item.

Software configuration item overview#

The software configuration item is the eo_data_embedding Python package together with its online documentation. Configuration management is handled entirely through git (source control and tagging), Semantic Versioning and pyproject.toml / __version__; a separate, standalone SCF artifact is therefore tailored out and is not part of the SDP §3 document tree. The remainder of this page records the inventory, baseline documents and means necessary for the configuration item, and cross-references the Software release note for differences from previous versions and the status of problem reports, change requests and waivers.

Inventory of materials#

The software configuration item is delivered in a Git repository on the EOPF. The software configuration item is constituted of all files contained on the main branch of the Git repository, comprising the eo_data_embedding Python package source, its pyproject.toml packaging metadata, the test suite and the online documentation. There are no additional binary or data elements: the library carries no bundled model weights or datasets, which are obtained at run time as described in the Software installation manual.

Baseline documents#

The following documents are included in the online documentation, and together with this SCF constitute the documentation applicable to the delivered software configuration item version:

Inventory of software configuration item#

The content of the software configuration item is the full contents of the main branch of the EOPF Git repository: the eo_data_embedding Python package, its packaging metadata (pyproject.toml), the test suite and the online documentation listed under Baseline documents. The released version is identified by the corresponding git tag and the package __version__ (Semantic Versioning).

Means necessary for the software configuration item#

eo-data-embedding is a pure-Python library. To develop, build and run the software configuration item the following are required but are not themselves part of the configuration item: a CPython interpreter and a PEP 517 build front-end (e.g. pip / build) driven by pyproject.toml; the third-party Python dependencies declared in pyproject.toml (notably PyTorch, the Clay encoder, FAISS, scikit-learn, pandas/pyarrow and NumPy), which are resolved and installed from the package index; and, for the GPU-oriented encode pass, a CUDA-capable runtime. The exact build and dependency-resolution process is described in the Software installation manual.

Installation instructions#

Please see the Software installation manual. There are no version-specific installation steps for this configuration item; the general installation procedure applies to every released version.

Change list#

Please see the Software release note for the changes incorporated into this version of the software configuration item.

Auxiliary information#

No auxiliary information beyond the inventory and means described above is required to characterise this configuration item.

Possible problems and known errors#

Please see the Software release note for the known errors concerning this software configuration item version.