MapInf partitions PWMaps

[Kalashnikovni]

Executing the following test:

mapInf(<<{[109:998]} -> |x-109|, {[108:108]} -> |x-108|, {[107:107]} -> |x-107|, {[106:106]} -> |x-106|, {[105:105]} -> |x-105|, {[104:104]} -> |x-104|, {[103:103]} -> |x-103|, {[102:102]} -> |x-102|, {[101:101]} -> |x-101|, {[100:100]} -> |x-100|, {[0:99], [999:999]} -> |x|, {[2000:2999]} -> |x-2000|, {[1000:1999]} -> |x-1000|, {[4000:4999]} -> |x-4000|, {[3000:3999]} -> |x-3000|, {[5000:5999]} -> |x-5000|, {[6990:7879]} -> |x-6990|, {[6100:6989]} -> |x-6100|, {[6000:6099]} -> |x-5900|>>);

generates a resulting PW of thousand maps, taking more than a second and a half to compute its result. Clearly no such partition should occur. I think that the problem is that we're performing composition, while the intention of mapInf is to calculate the result without using composition.

[Kalashnikovni]

With the new version it calculates the result in constant time, but it doesn't return 0 for every element in the domain.