At Thu, 25 Aug 2011 12:08:42 +0200, Štěpán Němec wrote: > > On Thu, 25 Aug 2011 06:25:29 +0200 > David Maus wrote: > > > As for the macro: What stop me to implement a macro for the generic > > operation is that for now the macro would depend on the global > > customization variable. That's not a problem per se but according to > > my readings about macros (mostly in context of Common Lisp, but that > > shouldn't matter) it should be considered bad style. > > Could you expand on this a bit? As far as I can tell, you obviously > shouldn't depend on a customisation variable at macro expansion time, > but I don't see how depending on it at run time is any different from a > function doing the same. > At expansion time the macro performs a transformation of the lisp at the place where the macro is used. At runtime the code of the expanded macro runs in the scope of the function where the macro was expanded into. A macro that uses a variable inside the expanded code that is not under its control (e.g. part of the argument list or gensym'd) is prone to introduce a bug caused by expanding the macro in an environment where this variable is not bound or used with a different semantics. In this particular case this should not be a problem indeed because we use a global dynamically scoped customization variable. Thus, whereever we would use the macro we can be sure the variable in the macro expansion is bound and carries the same meaning. Best, -- David -- OpenPGP... 0x99ADB83B5A4478E6 Jabber.... dmjena@jabber.org Email..... dmaus@ictsoc.de