Jonathan M Davis wrote: > > bearophile wrote: > > > Sean Kelly: > > > > std.paralellogrithm ;-) > > > > > > The module name I like more so far is the simple "parallel_algorithm". > > > But I don't mind the "p" prefix for the parallel function names. > > > > > > Bye, > > > bearophile > > > > It is also possible to have both "p" prefix and identical names. The latter > > would be mere aliases and would be activated by a version declaration or > > similar. I think both possibilities have their merits, so letting the user > > choose would be an option. > > > > parallel_algorithm is to the point, but not very concise. I do not think we > > can do very much better. > > That doesn't work. As soon as the non-p versions exist, you have name clashes. > So, if the point of starting the function names with p is to avoid name > clashes, then you've gained nothing. Besides, we generally try to avoid > aliases like that in Phobos. Such a scheme would never be accepted. > > - Jonathan M Davis
I think the point of having the same names is that you can parallelize code easily that was sequential before (or even for mere cosmetic reasons, by substituting the import, so no name clashes) But if there is only one option, then the "p" prefix is as far as I can see to be preferred, users can always do the aliases manually. Timon
