Loading ...
Sorry, an error occurred while loading the content.

80439Re: [XP] Re: Necessary comments?

Expand Messages
  • Robert Blum
    Oct 2 6:30 AM
    • 0 Attachment
      Hi Dale!

      > Another thought: Could you implement callChain in a
      > stack-friendly way?

      I think there's a misunderstanding. The call stack lives on the DMA
      processor. The only stack space is the return address, but there's a
      (very tight) limit on that.

      > And another: If appendChain works, that must mean that it causes
      > the commands to be called without overflowing the stack.

      Yes - the chain in question is physically copied behind the last
      command in the master chain.

      > So
      > could you force users to call appendChain instead of callChain?

      Not really, because I don't know at runtime how deep the calls are
      nested already. So, most of the time, callChain is the desired
      function. I /want/ to use it in most of my code.

      > Or rename callChain with some private name, then rename
      > appendChain as callChain? The underlying idea here is to move
      > the stack-chewing code to a method that users can't call in
      > limited-stack situations.

      The method stack space is not the issue - sorry I wasn't clear about

      - Robert
    • Show all 19 messages in this topic