Known differences between versions

Once in a while, a new version of PROJ causes changes in the existing behavior. In this section we track deliberate changes to PROJ that break from previous behavior. Most times that will be caused by a bug fix. Unfortunately, some bugs have existed for so long that their faulty behavior is relied upon by software that uses PROJ.

Behavioural changes caused by new bugs are not tracked here, as they should be fixed in later versions of PROJ.

Version 4.6.0

The default datum application behavior changed with the 4.6.0 release. PROJ will now only apply a datum shift if both the source and destination coordinate system have valid datum shift information.

The PROJ 4.6.0 Release Notes states

MAJOR: Rework pj_transform() to avoid applying ellipsoid to ellipsoid transformations as a datum shift when no datum info is available.

Version 5.0.0

Longitude wrapping when using custom central meridian

By default PROJ wraps output longitudes in the range -180 to 180. Previous to PROJ 5, this was handled incorrectly when a custom central meridian was set with +lon_0. This caused a change in sign on the resulting easting as seen below:

$ proj +proj=merc +lon_0=110 <<EOF
-70 0
290 0
EOF
20037508.34     0.00
-20037508.34    0.00

From PROJ 5 on onwards, the same input now results in same coordinates, as seen from the example below where PROJ 5 is used:

$ proj +proj=merc +lon_0=110 <<EOF
-70 0
290 0
EOF
-20037508.34     0.00
-20037508.34    0.00

The change is made on the basis that \(\lambda=290^{\circ}\) is a full rotation of the circle larger than \(\lambda=-70^{\circ}\) and hence should return the same output for both.

Adding the +over flag to the projection definition provides the old behavior.

Version 6.0.0

Removal of proj_def.dat

Before PROJ 6, the proj_def.dat was used to provide general and per-projection parameters, when +no_defs was not specified. It has now been removed. In case, no ellipsoid or datum specification is provided in the PROJ string, the default ellipsoid is GRS80 (was WGS84 in previous PROJ versions).

Changes to deformation

Reversed order of operation

In the initial version of the of deformation operation the time span between \(t_{obs}\) and \(t_c\) was determined by the expression

\[dt = t_c - t_{obs}\]

With version 6.0.0 this has been reversed in order to behave similarly to the Helmert operation, which determines time differences as

\[dt = t_{obs} - t_c\]

Effectively this means that the direction of the operation has been reversed, so that what in PROJ 5 was a forward operation is now an inverse operation and vice versa.

Pipelines written for PROJ 5 can be migrated to PROJ 6 by adding +inv to forward steps involving the deformation operation. Similarly +inv should be removed from inverse steps to be compatible with PROJ 6.

Removed +t_obs parameter

The +t_obs parameter was confusing for users since it effectively overwrote the observation time in input coordinates. To make it more clear what is the operation is doing, users are now required to directly specify the time span for which they wish to apply a given deformation. The parameter +dt has been added for that purpose. The new parameter is mutually exclusive with +t_epoch. +dt is used when deformation for a set amount of time is needed and +t_epoch is used (in conjunction with the observation time of the input coordinate) when deformation from a specific epoch to the observation time is needed.

Version 6.3.0

projinfo

Before PROJ 6.3.0, WKT1:GDAL was implicitly calling --boundcrs-to-wgs84, to add a TOWGS84[] node in some cases. This is no longer the case.

Version 7.0.0

proj

Removed -ld option from application, since it promoted use of deprecated parameters like +towgs and +datum.

cs2cs

Removed -ld option from application, since it promoted use of deprecated parameters like +towgs and +datum.

UTF-8 adoption

The value of all path, filenames passed to PROJ through function calls, PROJ strings or environment variables should be encoded in UTF-8.