Another small (No, really!) task for the interested

[Dan Sugalski]

As has been pointed out to me, we're falling really behind with our
embedding scheme. We have two *big* problems:

1) We've a sane, if unfinished API but no tests (besides Ponie, which
is a bit big) to make sure we actually implement the routines right
and don't mess things up.

2) We're not prefixing a lot (far, far too many) of symbols that we
really ought to be prefixing. (And we've got some symbols which only
differ in case, which is Really Bad)

So, the tasks.

We could really use some source tests to test the embedding interface.

We need to fill in the autoprefix.h header. (What header you say?
Update CVS! :)

This is a partial stopgap measure. What we *really* want to do is to
have the linker only expose those symbols we want exposed and hide
the rest, but that doesn't work everywhere, which means *all* our
non-statics need to be prefixed. Hence the autoprefix.h file. This is
where all the "#define foo(x,y) Parrot_foo(x,y)" defines need to go.

Anyway, what we need to do is figure out what symbols we're exposing
that aren't prefixed and get entries for them in here. There's a tool
(tools/dev/nm.pl) that works many places that gets a list, which'd be
a good place to start.

Not thrilling work, but not bad either as these things go. Could be
worse, certainly.
--
Dan

--------------------------------------it's like this-------------------
Dan Sugalski even samurai
d...@sidhe.org have teddy bears and even
teddy bears get drunk