tberghammer added a comment. I am not sure if doing all of the parsing lazily is the good approach because of speed considerations.
The problem is that if we do everything lazily then in one side we are parsing only the information what we need but on the other side it is very difficult (or impossible) to make it multi threaded. In case of a high end machine I think doing the parsing in 8-16 core (or even more) can be faster then the lazy approach if we have to parse a lot of information anyway. I don't have any measurements but looking into the direction we are heading now we will need to parse more and more information to improve the expression evaluation (e.g. to support limit debug info) and to get it a parse a lot of thing upfront might be a better approach. In terms of the performance I expect it to be a small to medium difference (not sure about memory) but will speed up the debugging experience later. For me it is more annoying when I have to wait to evaluate an expression then to wait for the debugger to start up. Repository: rL LLVM http://reviews.llvm.org/D15312 _______________________________________________ lldb-commits mailing list lldb-commits@lists.llvm.org http://lists.llvm.org/cgi-bin/mailman/listinfo/lldb-commits
