rgdal 1.4-2 on CRAN

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

rgdal 1.4-2 on CRAN

Roger Bivand
The latest release of rgdal is the first step towards support for PROJ 6
(PROJ is the generic name for proj4). From PROJ 5, some internals were
changed but the API and metadata files remained unchanged. PROJ 6,
released 1 March, requires software to declare that it uses the old API,
which rgdal still does, but changes the metadata files to better
accommodate the changes in the internals. These are stepping away from the
+datum= tag, and there are now some coordinate reference systems (CRS)
that cannot be represented using proj4 strings. PROJ 7 is due in March
2020, and will remove the old API.

The sp and rgdal packages will continue to use the proj4 string
representation of CRS; sf also permits the use of EPSG codes, so
(potentially) permitting non-proj4 string CRS to be handled. rgdal and sf
will migrate to the new PROJ API during 2019 and early 2020, but will
attempt to maintain backward compatibility for more recent releases of
PROJ. Why the changes? Well, WGS84 was from 1984, and the world has
changed quite a lot since then, so pivoting CRS to CRS transformation
through geographical coordinates using the WGS84 datum is becoming more
error-prone. The new PROJ internals pipeline transformations without
pivoting through WGS84, so enhancing precision and making calculations
(more) future-proof.

I've notified the maintainers of several affected packages so that they
have time to adapt before PROJ 6 propagates to upstream packagers (GRASS
already uses the new API and GRASS master is PROJ 6 ready, GDAL master is
PROJ 6 ready). We'll hold off putting PROJ 6 into Windows and OSX binaries
for some months, but please be ready us to do that, latest end of year
2019, to be ready for PROJ 7.


Roger Bivand
Department of Economics, Norwegian School of Economics,
Helleveien 30, N-5045 Bergen, Norway.
voice: +47 55 95 93 55; e-mail: [hidden email]

R-sig-Geo mailing list
[hidden email]
Roger Bivand
Department of Economics
Norwegian School of Economics
Helleveien 30
N-5045 Bergen, Norway