Image as database (was: Re: Minnow WIKI Migration)

Philippe Marschall philippe.marschall at
Mon Oct 23 20:55:53 UTC 2006

2006/10/23, Bert Freudenberg <bert at>:
> Am 23.10.2006 um 21:30 schrieb Philippe Marschall:
> > 2006/10/23, Cees de Groot <cdegroot at>:
> >> On 10/23/06, Philippe Marschall <philippe.marschall at> wrote:
> >> > > > So about 300 Euros?
> >>
> >> [...]
> >>
> >> > 64bit VM?
> >> >
> >> You pay the hosting bills for a new box? ;)
> >
> > I'm willing to pay 2 GB of RAM if that's what is needed to run Pier.
> > That Squeak can't handle this is a Squeak specific limitation that has
> > nothing to do with the point that memory is that cheap.
> > As pointed out numerous times on squeak-dev and disputed by none, all
> > VM related issues can be fixed easily by just fixing the VM. This is
> > no problem since the VM is open source.
> If we had a transactional virtual object memory that's continuously
> saved to disk (think OOZE/LOOM), that might be viable. Perhaps with
> Magma you could have almost the same semantics, just be careful what
> you touch. But not with the current object memory. No way. Not if you
> care about the data.
> It's not about RAM being cheap or not. It's about designing for the
> common and the worst case. Why you would want to bring in gigabytes
> of data if the working set is just a few megabytes is beyond me.

The point was just that holding the whole wiki in the memory is no
problem memory or money wise.

That the vm, like in many other cases too, is the real problem (and
I'm quite sure the Java VM would be up to it) is a completely
unrealted issue.


More information about the Squeak-dev mailing list