LoginSign Up
← Back to all sparks
H

hydroloom

ANALYTICS
Velocity0.0

R package hydroloom by doi-usgs — release notes from GitHub.

github.com/doi-usgs/hydroloom ↗

USGS puts a type system over its river network toolkit so errors surface at dispatch

hydrologynetwork-analysisgeospatialr-packageusgs
◆Current state
hydroloom builds and navigates hydrologic flow networks, carrying functionality migrated out of nhdplusTools. Version 1.2.0 introduces an S3 class hierarchy — hy_topo, hy_leveled, hy_node, hy_flownetwork — assigned automatically by hy() and by producer functions, letting the package validate input at dispatch time and emit guided errors. Outlet detection is now defined explicitly: a row is an outlet when its toid is not in id, with reserved values, NA and implicit absence all accepted.
◆Where it's heading
The package spent its first releases porting and broadening — non-dendritic network support, divergence routing, subsetting that follows diversions out of a basin — and has now turned to making that surface safe to use. The class hierarchy is the structural expression of that turn: instead of every function re-checking whether a data frame has the columns it needs, the type carries the guarantee. The explicit outlet rule resolves a category of failure where valid networks errored on NA or orphan toid values.
◆Prediction
The release notes flag that subclass attributes are stripped by standard dplyr operations, which is the kind of rough edge that usually generates follow-up work — expect attribute preservation or restoration helpers next.

◆Recent moves

  1. 2mo ago

    hydroloom v1.2.0

    ⚡ SPARK

    An S3 class hierarchy — hy_topo, hy_leveled, hy_node, hy_flownetwork — is stamped automatically by hy() and by producer functions, enabling dispatch-time validation and guided error messages. Outlet detection becomes an explicit rule rather than an inference, accepting reserved values, NA, implicit absence and unique-per-outlet ids; calls that previously errored on NA or orphan toid now succeed. Existing data.frame and hy code keeps working.

    View source ↗
  2. 5mo ago

    Test tolerances relaxed for CRAN Fedora checks

    Loosens test tolerances so CRAN's Fedora checks pass. Infrastructure only.

    View source ↗
  3. 5mo ago

    Network subsetting and divergence-routed accumulation

    subset_network() arrives with support for including diversions emanating from a basin, accumulate_downstream() gains total-upstream and divergence-routed modes, and make_index_ids() is rewritten with to, from and both modes — deprecating make_fromids() and format_index_ids() in the process.

    View source ↗
  4. 10mo ago

    Sort and indexing fixes

    Bug fixes for sort_network with duplicate extended attributes, plus id-specific search in index_points_to_lines and error handling.

    View source ↗
  5. 1y ago

    Upmain and downmain navigation for non-dendritic networks

    Non-dendritic support deepens: upmain and downmain navigation in navigate_network_dfs() and make_index_ids(), a to_flownetwork() representation, and navigation that now follows a diverted path back to a main path rather than only traditional tributaries.

    View source ↗
  6. 2y ago

    Initial release completing the nhdplusTools migration

    The founding release, completing the migration of network functions out of nhdplusTools: the hy() S3 object, depth-first navigation, non-dendritic sorting, stream order and stream calculator attributes, linear indexing without measure attributes, and support for both NHDPlus and NHD 24k naming.

    View source ↗