Michael wrote: > This and especially portability is why I argued-- > unsuccessfully--
? I thought the question was still up in the air for g7 and for now it was user choosable by the way you constructed the db.connect string. > for a separate sqlite DB for each vector file, even if it is > inefficient. is it? only when doing joins? if so, how common a use case is that? (no idea how well sqlite handles fseeks() to jump right to the data it needs, or if it has to read in the whole file) I'm mildly in favour of per-map dbs as I've often seen 600mb vector datasets on 32bit machines, and 2-4gb limit of the filesystem doesn't seem so far away if those get set to be cumulative. But I am fairly ignorant of DB issues and sqlite implementations in particular, so am not really familiar with what gains you might see from the monolithic approach. best, Hamish _______________________________________________ grass-user mailing list [email protected] http://lists.osgeo.org/mailman/listinfo/grass-user
