From: ohland@charter.net   
      
   MAJ Tom, here is a small pothole on the Golden Path.   
      
   Though I'd expect RLE to be an optional implementation. Maybe, maybe   
   not. YMWV...   
      
      
   As briefly discussed earlier, the ECP Protocol includes a Simple   
   Compression Scheme called Run Length Encoding. It can support a maximum   
   compression ratio of 64:1 and works by sending repetitive single bytes   
   as a run count and one copy of the byte. The run count determines how   
   many times the following byte is to be repeated.   
      
   For example, if a string of 25 'A's were to be sent, then a run count   
   byte equal to 24 would be sent first, followed by the byte 'A'. The   
   receiving peripheral on receipt of the Run Length Count, would expand   
   (Repeat) the next byte a number of times determined via the run count.   
      
   The Run Length Byte has to be distinguished from other bytes in the Data   
   Path. It is sent as a Command to the ECP's Address FIFO Port. Bytes sent   
   to this register can be of two things, a Run Length Count or an Address.   
   These are distinguished by the MSB, Bit 7. If Bit 7 is Set (1), then the   
   other 7 bits, bits 0 to 6 is a channel address. If Bit 7 is Reset (0),   
   then the lower 7 bits is a run length count. By using the MSB, this   
   limits channel Addresses and Run Length Counts to 7 Bits (0 - 127).   
      
   Louis Ohland wrote:   
   > Fortified with Caffeine and an insatiable desire, I'm looking at the ECP   
   > and Type 3 information [such as it is] with the goal of enabling some   
   > sordid ECP to DMA port functionality.   
   >   
   > Gonna have to drop it down into granny low gear and slowly let out the   
   > clutch...   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)