4 ms·
Given that type inference can be expressed in the form of constraints, could this be used to speed up type inference on huge programs?
by Apanatshka 11y ago
Given that type inference can be expressed in the form of constraints, could this be used to speed up type inference on huge programs?
- coolsunglasses 11y agoCould be used for a package dependency solver like Cabal's, possibly.
- dhaivatpandya 11y agoIt could be, but the primary issue with the Communication Free Learning algorithm is the fact that it is very hard to tell when a given constraint set has no feasible solution by directly trying to solve it. That's because Communication Free Learning is a probabilistic algorithm that is meant to converge to a solution over time. But, if after say 10k iterations we still haven't found a solution, that could mean that there isn't one or that we just haven't stumbled on it yet. However, it is possible to apply the algorithm in situations where we know that there is a solution, at least with a pretty high degree of confidence.