Absolute path in TGIS connection can make db inaccessible in a (mixed) multi-user environment

[sbl]

It seems that the connection to the TGIS sqlite.db is defined for the mapset using an absolute paths.

That makes it inaccessible in cases where mount points change, e.g. in a multi-user environment. In my particular case, /grassdata/ETRS_33N/precipitation_daily/tgis/sqlite.db (produced from within Linux) is (naturally) not readable from Windows where the equivalent path would be R:\grassdata\ETRS_33N\precipitation_daily\tgis\sqlite.db

One can work around this issue using t.connect -d when starting to work, but this is only possible for "mapset owners".

Would be nice if environment variables could be used ($GISDBASE/$LOCATION_NAME/$MAPSET) and first expand/evaluate at access time or relative path to the current mapset...

[neteler]

Does the TGIS sqlite.db path implementation differ from the vector sqlite.db path implementation?

[huhabla]

Replying to sbl:

It seems that the connection to the TGIS sqlite.db is defined for the mapset using an absolute paths.

That makes it inaccessible in cases where mount points change, e.g. in a multi-user environment. In my particular case, /grassdata/ETRS_33N/precipitation_daily/tgis/sqlite.db (produced from within Linux) is (naturally) not readable from Windows where the equivalent path would be R:\grassdata\ETRS_33N\precipitation_daily\tgis\sqlite.db

One can work around this issue using t.connect -d when starting to work, but this is only possible for "mapset owners".

Would be nice if environment variables could be used ($GISDBASE/$LOCATION_NAME/$MAPSET) and first expand/evaluate at access time or relative path to the current mapset...

This issue should be solved in 7.2 release branch and trunk. The connection should use relative paths to the tgis database as default. However, be aware that the temporal framework was not tested on windows and is not designed to run on windows systems.

[sbl]

Given that we are relatively close to the release of of 7.2 and that this is probably an edge-case anyway, I suggest to close this issue as fixed.

BTW: At least point-sampling of STRDS worked fine for me on Windows, which is probably what most of our Windows users would want to do...

[sbl]

Feel free to reopen if needed in 7.0.x