experiments with clipboards and clipboard managers

[Ben Pfaff]

I did a little bit of research tonight after thinking about our
IRC conversation about when it takes a long time to compose
selection data for a clipboard response.  I modified the function
that responds to a clipboard request to printf a message, and
then tried a few clipboard copies with two different clipboard
managers running (Parcellite and Diodon) and with no clipboard
manager running.

Parcellite's behavior surprised me.  Twice a second, it requests
a new copy of the primary selection from whatever app owns it.
If an app has a big selection, it essentially DOSes the machine.

Diodon seems a little better.  It appeared to request the
selection three times (three formats? didn't check) and then
stops.

Without a clipboard manager, there's no problem.

Gnumeric doesn't handle large copy/cut/paste particularly well
either.  It's easy to induce long pauses (seconds, maybe tens of
seconds?) with a few hundred thousand cells.