From: nowhere@never.at   
      
   Kerr Mudd-John wrote:   
      
   >>> You inspired me to think of a horrible^w brilliant scheme, but this   
   >>> margin is too small to contain it.   
   >>Let me guess what you had in mind...   
   >> me too tried to get rid of the 2:1 at all and create the 4:3 decoder   
   >> with a 1:1 string and 'some' SMC modifications to fit into 041..07f.   
      
   > Yup; a whole heap of mods to the first line to make it a b64 decoder.   
      
   I'm already (on paper only yet) down to 7 single and 6 paired modifiers   
   by saving two singles on  66 C1 05 '46' and 66 C1 2C '48' instead of '06'   
   and '08'. This still needs 3*13+4 bytes for modification and a 59 byte   
   string, not sure yet if I can make this any shorter.   
      
   The first line may just need to create a backjump and jump the second   
   line where a REP MOVSB (to 0x44) and a forward branch are created in   
   the first line. And if the jumping RET can't be created with [bx+..],   
   we can use [bx+di+..] (0100+0044+..).   
   ...   
   > I'm sort of working on it; I've been away from the computer last week.   
      
   me too work on it, but only during rare breaks.   
      
   >> btw: you dont need [BP+SI] [BP+DI] at all:   
   >> we can use [32 34,30 34] and [32 35,30 35} instead (BP is free then)   
      
   > that scheme was all about using an extra reg to cmp; with the only one being   
   > ch from 2B   
      
   > 38 2B          cmp          [bp+di],ch   
      
   > it saved a byte or 2, I think.   
      
   I see, it saved one byte here but needed SI<>DI swap.   
   __   
   wolfgang   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)