force_FC = TRUE no longer adds a rmapshaperid id column. force_FC
is also better respected across functions and methods; previously it was
somewhat inconsistent (#151).gj2008, which if TRUE outputs geojson following the old
(2008) specification. Polygon rings are CW and holes are CCW, which is the
opposite of the default RFC 7946-compatible output. This is required by D3, and
packages that depend on it, like plotly (#167, #168).This is a fairly major release with much of the internal plumbing changed. I have tried to keep user-facing changes to a minimum, but please report any issues to https://github.com/ateucher/rmapshaper/issues.
geojsonsf package instead of geojsonio for object conversion (#118).v0.6.25 (#130).geojson_list objects. This was a rarely-used class from the geojsonio package (#118).force_FC, sys, and sys_mem are now passed to apply_mapshaper_commands via ... rather than explicitly, so they are now documented in the ... section of each function. This may break some existing code if you were passing values to these arguments by position rather than by name, especially using force_FC in ms_simplify as it was not at the end of the argument list. It may also change the class of the return value for some input classes and functions (such as ms_lines and ms_innerlines) as force_FC will inherit the default TRUE for all functions.quiet argument to silence mapshaper console messages when using sys = TRUE. This can be controlled globally with options("mapshaper.sys_quiet" = TRUE) (#125).options("mapshaper.sys_mem"=X), where X is the amount of memory in GB.units columns in sf objects would cause failures; all numeric columns of class "units" are now converted to numeric before running through mapshaper commands. (#116, thanks @Robinlovelace)force_FC in apply_mapshaper_commands(). The default value is TRUE (#120, thanks @dblodgett-usgs)check_sys_mapshaper() - fixed description of return value (#117, thanks @dblodgett-usgs).tmpdir()
path and sys = TRUE (#107)ms_erase (#110, #104, #112)sys = TRUE, now uses mapshaper-xl in the system call, allowing
larger memory use. Default 8GB can be specified in new argument sys_mem (#94, #112)system2() over system() for flexibilitylibv8 installed, as they do not support many
aspects of modern JavaScript (ES6). This appears to only impact ms_erase()
and ms_clip().apply_mapshaper_commands() no longer deletes a file when used on a local file (#99, #100)rgdal to Suggests so ms_clip and ms_erase can transform
Spatial* objects when they have different CRSs.sf object is a tibble, the output is now now also a tibble,
(#95, thanks @mdsumner)mapshaper v0.4.107sys = TRUE would fail on Windows in some circumstances (#77)rmapshaper functions on sfc objects failed with
sf v0.7sys argument to allow the use of the system mapshaper if it's installed (#61)mapshaper v0.4.64 (#60)sf::st_read() is now used throughout for reading from disk and from geojson
strings, which allows for greater consistency and better performance.rmapshaperid column is only retained if it is the only column, otherwise it's dropped.ms_innerlines returns only the geometry for sf and Spatial*DataFrame classes. (#57)ms_dissolve gains a weight argument for generating weighted centroids of dissolved points. (#39)snap_interval to ms_simplify() (#43, @nikolai-b)drop_null_geometries argument in ms_simplify.geo_list() (#45, @nikolai-b)keep in ms_simplify() were converted to scientific notation (#48)weighting argument to ms_simplify() (#27)remove_slivers argument in ms_clip() and ms_erase()