> As I understood the discussion at the last VXL meeting, we agreed that we
> should port the algorithms in tbl into vil/algo
Sure, I have no problem with that.
(But, for the time being, most of the algorithms in vipl have not been
ported, and it's still much easier to implement a vipl algorithm than
a vil algorithm.)
> the vipl interface is very daunting - too many templates.
But that is what makes it so powerful.
With just a few new files (see tbl/vepl2/accessors) I was able to "port"
vipl to vil2. Similarly, anybody can use vipl algorithms on their own
> vipl leaks a lot of memory, despite your and my attempts to fix that.
That is correct.
The vipl_section_container etc. objects should be made smart pointers.
> There isn't any problem with continuing to work on vipl, etc. but that isn't
> the direction vxl is heading at the moment.