requirements.txt

# Flat-list runtime dependencies for Streamlit Cloud.
#
# Streamlit Cloud reads requirements.txt (not pyproject.toml extras),
# so we keep this in sync manually with [project] +
# [project.optional-dependencies.viz, .dashboard] in pyproject.toml.
#
# We DO ship cartopy here even though it's the slowest install in the
# tree (~25-30s cold start) — the dashboard's 3D globe reads coastlines
# from cartopy's bundled Natural Earth shapefiles via
# visualization/figures.py:_add_coastlines. Without cartopy the 3D
# Earth becomes a featureless blue ball and the "looks like Earth from
# space" hero falls apart. system-level libgeos-dev + libproj-dev
# come from packages.txt.
#
# The trailing `.` installs the project itself (sat_tracker package
# from src/-layout). Without that line, `from sat_tracker...` imports
# inside the dashboard would fail with ModuleNotFoundError on the cloud.
#
# Dropped vs. [dev]: pytest. The cloud runtime never runs tests.
# Dropped vs. [viz]: matplotlib (the dashboard never uses the
# static-cartopy renderer; only the deferred shapefile-reader path).
sgp4>=2.23
skyfield>=1.46
requests>=2.31
numpy>=1.24
plotly>=5.18
kaleido>=0.2,<1.0
streamlit>=1.30
cartopy>=0.22
.