[Vm-dev] Eliot's BlockClosure model questions

[tim Rowledge]

On 29-07-2013, at 1:28 PM, Clément Bera <bera.clement at gmail.com> wrote:
>  Actually on second thoughts it may be that blocks need the temp vector to work with the stack mapping: if a block outlive its homeContext, the homeContext variables may be overridden in the stack.

No, VW used stack mapping of various sorts a long time before that. It was just a fair bit more complicated to handle returns IIRC. I think the issue was that making sends a tiny bit slower (to handle the temp array) could make a large number of expensive returns much faster and thus improve overall performance quite a bit.

tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
Security announcement - as of next week, passwords will be entered in Morse code.