What are the advantages and disadvantages importing vs linking data in GRASS?

Question

I'm fairly new to GRASS, and only recently discovered the r.external and v.external functions.

I'm used to working with R and QGIS where all the data is "external". Typically I keep vector data in PostGIS and rasters in the filesystem. But it seems the GRASS paradigm is to import this data into mapsets (with r.in.* and v.in.*).

So, what are the advantages of switching to that paradigm, vs. using r.external/v.external? Is one or the other faster, or easier to keep organized, or more bug-prone or error prone? Are there formats that can't be handled by linking?

Answer 1

GRASS 6.x:

vectors (v.external()):

• they are virtually linked into the current mapset in read-only mode.

from v.external():

The simple feature data model used by OGR is very different from the topological format used by GRASS. Instead of true topology, so called 'pseudo topology' is created for data linked by v.external. User should learn the difference between those to formats, because some modules working correctly with GRASS native data, can produce wrong results with input layers created by v.external.

rasters (r.external)

• r.external just register the files rather than physically importing them. The native GRASS format (r.in.gdal import) has many advantages for processing.

I use them only as background maps.

GRASS 7

• supports write access (working with external data in GRASS 7)