This is the mail archive of the guile@cygnus.com mailing list for the guile project.
Index Nav: | [Date Index] [Subject Index] [Author Index] [Thread Index] | |
---|---|---|
Message Nav: | [Date Prev] [Date Next] | [Thread Prev] [Thread Next] |
>> It seems to me you've just identified a classic case where multi-methods > are >>great. > >Yes, that was the point. Of course that assumes you're satisfied >with hand-waving. But I want an engineering design (pattern?) >for a maintainable, extensible, robust understandable package. >I want someone to show me how multi-methods makes that easier. > If you can't get the built in dispatch to do what you want then MOP? > >If you suggest using the MOP then you're basically agreeing that multi-methods are >not useful for real systems. I'm no expert on MOP, but it seems to me that multi-methods, with or without MOP is a way to formalise and standardise a way to define functions that work differently according to their type. If I've got 1000 functions to write for different display devices and display objects, do I really care if I have to fiddle with the MOP in order to get the dispatch algorithm right according to how I want things done? Not that I can really see why you need to change the MOP anyway. The point of multi-methods is that the 1000 functions can be fully independant. I can just write it and load it. I don't have to worry about where it should be called from, or how to dispatch to it. Isn't that self-evidently a good thing?