On Sun, Jun 1, 2008 at 12:15 AM, Felix Rabe
<[EMAIL PROTECTED]> wrote:
> - JIT.  I think it was Ian Piumarta who said in his recent presentation
> in Germany that nowadays processing power is cheap.  So why keep
> software around in binary blobs anyway?  As I see it, the computer
> should be able to decide what software should be kept around in a
> (just-in-time and then cached) compiled state and what should not.

Marcus Denker's PhD is just about that, keeping code in AST form and
using the compiler as an optimizer/instrumenter and bytecodes as a
cache.

> - Memory management.  I hate the save button.  Why do I have to tell a
> computer that a string of bytes in one kind of memory should now move to
> another kind of memory that can *also* handle a string of bytes, but
> just possesses a different set of attributes?  (Like: disks are larger

Well the save button is also kind of a commit, or checkpoint, that you
can use for large-grained undo. In a fully-persistent world you would
probably want to extend it to some kind of versioning system where the
save button means "this is in a meaningful state".

> long-term preservation.  In the system I envision, a power failure could
> maybe result in something like the last 10-30 seconds of my work lost,
> but never more, as memory portions would be constantly swapped into and
> out of disk in a safe (journalled?) fashion, in a way that would not
> interfere with the user-visible performance of the system.  Of course,

You could also just journal every user action and replay them from the
latest full snapshot in case of a problem.

> sensible and usable than a boot logo showing up.)  If OLPC can put an
> XO-1 to sleep and wake it up again within a fraction of a second, then
> the shutdown-bootup cycle should be trimmable to a similar order of
> magnitude with a future operating/computing system, I guess.

Well this is waking up from sleep, not booting from a clean image. The
apple laptops are reasonably fast at waking up too.

-- 
Damien Pollet
type less, do more [ | ] http://people.untyped.org/damien.pollet

_______________________________________________
fonc mailing list
[email protected]
http://vpri.org/mailman/listinfo/fonc

Reply via email to