I'm following this list only for a very short time now, and I don't have very much programming experience, so tell me if all this is pure nonsense.
But isn't this partly addressed with COW? The way I understood Copy On Write, it means that if you "copy" a string, only some pointers are copied and some flags are set, until you modify one of the two strings. So IMHO, this shouldn't be too much slower than simply copying the pointer of an immutable string. And if you wanted to change a mutable string buffer (which has been passed to your code by reference, for example), you might want to make a copy anyway, which is exactly what happens when you modify a "copied" string in Parrot. Thus I don't see a very big performance win in introducing mutable strings. Am 10.07.2002 14:00:16 schrieb John Porter <[EMAIL PROTECTED]>: > >"Clark C . Evans" wrote: >> So, I propose that Parrot supports both of these attitudes >> of reasoning by using a flag. Initially, all strings >> are 'mutable'. > >I concur with Clark. What he's proposing is (it seems to me) >nothing more than a "read-only" flag, and that's as >immutable as a variable needs to get. > >However, I'm concerned about the performance impact of such a thing. >Perhaps a better approach would be to swap out the mutator methods >with code that throws. > >-- >John Douglas Porter > > >__________________________________________________ >Do You Yahoo!? >Sign up for SBC Yahoo! Dial - First Month Free >http://sbc.yahoo.com > >
