Re: dynl.c

[Marius Vollmer <mvo at zagadka dot ping dot de>]

Dirk Herrmann <dirk@ida.ing.tu-bs.de> writes:

> A first start made me look at dynl.c.  I enclose a proposed patch,
> but would be grateful if someone with better knowledge of the
> workings of dynamic loading could take a look at it.

It looks basically good to me.  I'm unsure about the wording of the
new error messages, tho.  I think they could be changed to

    "Already unlinked: ~S"

> However, I do have a question about the implementation: Why is the
> user required to explicitly unlink a library?  Couldn't this be done
> automatically by the gc?

Ideally, the GC would do this, but I'm not sure whether the GC has
enough knowledge about the use of a dynamically linked object.  FOr
example, you can't just unlink a shared object when the smob for it is
collected.