From an API level I do not think there is a big deal. The resolver
could just fetch all resources at start. It can of course only return
a single solution. This might be unfortunate but I find it hard to see
why that is a limitation since any solution that satisfies all
requirements should be
Hi guys,
Some documents about the way the SAT encoding works are available here:
https://hal.archives-ouvertes.fr/hal-00870855/document
and here:
http://dev.eclipse.org/mhonarc/lists/p2-dev/pdf7BeCW3Xx01.pdf
If you are willing to give a try to a SAT encoding of uses, I am happy
to give a hand.
I remember trying to map uses constraints to a boolean expression but could not
find any way that did not blow up the expression size. This seemed very
unfortunate because I think they can actually be used to reduce the search
space considerably.
From an API level I do not think there is a big
I will be interested to see if you can successfully map the OSGi uses
concept into the SAT solver p2 uses. I briefly looked at that a long time
ago when we were refactoring the Equinox framework (Luna) and were
replacing the old Equinox resolver. It was far from obvious how you would
achieve
