If memory serves me right, Dan Sugalski wrote: > Currently open is the situation of flags and such from more complex > calls. (Like what we do if we get back a pointer that's getting > stuffed into a PMC--do we set the type, if so what type, and what do > we set the flags to?)
So is it totally type unsafe ? ... (as you know I'm obsessed with type safety) ... "pMYPMC;pp" Does this appeal to anybody .... [Disclaimer: idea borrowed totally from JVM signatures "Ljava/lang/String;(Ljava/lang/String;i)"... Where the Long 'L' was later on modified to carry type info of the pointer...] > I'm thinking, as part of this, that more introspection on buffers and > PMCs is in order, so we can have bytecode manipulate their flags and > types properly. Well that does not totally satisfy me .... I might be expecting a XPMC when the pointer is stuffed into a YPMC and my program blows up in the middle .... (of course throwing an exception is not technically blowing up, it's damage control).. But well ... I should admit , I'm just trying to make the C# to parrot compilation as safe as IL .... (PInvoke and stuff...) Gopal -- The difference between insanity and genius is measured by success
