[PATCH] D13193: Add an MCTargetMachine and use it to configure MCAsmInfo

[hfinkel@anl.gov via llvm-commits]

hfinkel added a subscriber: hfinkel.
hfinkel added a comment.

As part of this change, you're creating classes whose primary function seems to be to cache results from calling member functions of Triple? Is there some performance benefit to doing so? Maintaining this pattern increases the maintenance burden on using Triple queries; why not just expose the Triple member for querying?


http://reviews.llvm.org/D13193