At 11:29 AM 10/23/2001 -0400, Gregor N. Purdy wrote:
>I'm wondering if there is something we can refactor so that we don't
>have to link interpreter.o into things like like pdump. Pdump doesn't
>want to be an interpreter. It doesn't want to run ops. But, passing
>the interpreter arg into PackFile_unpack means the caller needs to
>have one laying around, and that cascades into having to link a bunch
>of otherwise unnecessary stuff into pdump.
Yeah, that's because all the memory allocation stuff for perl-specific
things needs to vector off of the interpreter structure. I'm already
looking to do some refactoring, since I've become Officially Annoyed at
having to pass around interpreter structures. Bleah. Since we're
interpreter-per-thread I think I'm going to stow it in a TSD slot and not
pass around an otherwise useless argument.
As for detaching it from the interpreter, that's an interesting thing, but
I don't know that it'll work. To handle strings properly means potentially
dynamically loading string libs anyway, and those libs may well be written
in Perl. (Well, parrot bytecode, but still...)
Besides sheer size of executable (which we can get around by building the
core interpreter as a shared library) is there any other reason to not have
an interpreter structure hanging around?
Dan
--------------------------------------"it's like this"-------------------
Dan Sugalski even samurai
[EMAIL PROTECTED] have teddy bears and even
teddy bears get drunk
