#1020: Reduce WeylScal4 code size ------------------------------------+--------------------------------------- Reporter: eschnett | Owner: Type: enhancement | Status: review Priority: major | Milestone: Component: EinsteinToolkit thorn | Version: Resolution: | Keywords: ------------------------------------+---------------------------------------
Comment (by hinder):
At the moment, I am reluctant to support this patch. The expressions become less readable and are now different to the expressions in the original paper they were copied from, making them less verifiable. Erik's comment that his UseCSE solution doesn't look any nicer than the original applies only if you read all the code (i.e. the expression and the hint). In the new form, the actual expression is in the form that the reader expects, and might be checking from a paper. The CSE hint is something the user can ignore, as Kranc guarantees that this is only an optimisation, and not a change in functionality. I would like for the source expressions to be directly comparable with the paper they are based on.
The original reason for wanting to implement this change was due to a build failure on BlueGene/Q. Does the code still fail to build, now that some of the Riemann symmetries have been implemented? If it now builds successfully, then in my opinion it is better to have simpler, more readable Kranc source than to reduce the build time and size of the generated C++.