From: Martin A. <ma...@at...> - 2001-08-01 18:40:13
|
William Harold Newman wrote: > > On Wed, Aug 01, 2001 at 05:06:08PM +0200, Martin Atzmueller wrote: > > Attached is a patch for src/pcl/walk.lisp, and some test code for that > > (based upon the walker tests in walk.lisp): [...] > I reviewed it some. I'm not sure it's a perfect solution, since the > PCL approach of embedding everything in a function still seems a > little bizarre. I think someday it might be possible to do something > more straightforward, given that, unlike PCL, we don't need to be > portable across implementations. But it looks to me as though your > BOGO-FUNCTION type is distinctly cleaner than what the walker was > doing before. Yes, perhaps it's not the perfect solution, but it seemed like the best solution, if the walker-environment is still stuffing its info into the LEXENV structure. The approach is somewhat unusual, yes, but if it's done like that, SBCL's internals seem just to ignore the bogus-function-information. I thought about stuffing the list as-is in the lexenv structure, but this would have violated the "assertion" that the elements in the :function slot of LEXENV are like (<name> sb-c::macro <macro-function>), and so I wasn't sure, whether the compiler would like that. -- Martin Atzmueller <ma...@at...> |