The switch may have been a bad idea. It's not very intuitive. There was a reason behind it though. With the old queue based approach sometimes there was no way to get the tasks connected properly without using named pipes. With the stack based approach I think it's always possible to connect tasks without using named pipes.
Consider this example: I want to read two xml files, apply a change to both of them, then merge the two results into a single output file. With the old queue approach. osmosis --rx file1.osm --rxc change1.osc --ac --rx file2.osm --rxc change2.osc --ac --merge --wx out.osm This fails because the second --ac takes the first --ac result as the first input and the second --rx as the second input. I can't think of a way to re-order tasks to get around the problem. With the new stack approach. osmosis --rxc change2.osc --rx file2.osm --ac --rxc change1.osc --rx file1.osm --ac --merge --wx out.osm This works (I haven't tried it but I assume it does :-). So the stack based approach is more flexible, but takes a few more mental gymnastics to get it right. To be honest I haven't found it too bad one I got used to it. It was originally a suggestion by Martijn, but if the general consensus is that it's confusing I can revert it back again. Frederik Ramm wrote: > Hi, > > Karl Newman wrote: > >> I think it's because of a change to a stack-based command-line parsing (from >> a queue). Try swapping the position of the --rx and --rxc parameters. >> > > Works, thanks. A bit conter-intuitive if you ask me, perhaps I should > make it a habit to use named pipes... > > Bye > Frederik > > _______________________________________________ dev mailing list [email protected] http://lists.openstreetmap.org/listinfo/dev
