Re: binary interface

[Dirk Herrmann]

On Thu, 1 Mar 2001, Keisuke Nishida wrote:

> At Mon, 26 Feb 2001 12:27:25 +0100 (MET),
> Dirk Herrmann wrote:
> > 
> > > (Just a quick response.)  I want to load bytecode using mmap so that
> > > several processes can share the same memory.  Isn't it a good reason
> > > enough?
> > 
> > I don't understant that.  Why is sharing of memory dependent of the
> > question whether you choose a binary representation for storing?
> 
> Because that way we can use shared mmap:
> 
>   From `man mmap':
>        MAP_SHARED Share this mapping  with  all  other  processes
>                   that map this object.  Storing to the region is
>                   equivalent to writing to the  file.   The  file
>                   may  not  actually be updated until msync(2) or
>                   munmap(2) are called.

I think I understand the point.  However, is this actually a concern in
the current state of development?  It seems to me that the use (or
non-use) of mmap is an optimization that we can decide about later.  Thus,
we should not modify core parts of guile (like the implementation of
smobs) to prepare for such an optimization now.

However, we may find later that such a feature is interesting for guile.
Thus I suggest to either turn the current solution (and maybe parts of the
discussion) into a proposal, and/or tag the current state of cvs such that
we can resurrect the implementation at that time.  For now, we should not
add this stuff.  (The decision, however, is certainly up to the
maintainers.)

Best regards,
Dirk Herrmann