[Vm-dev] [Pharo-project] Plan/discussion/communication around new object format

[Eliot Miranda]

On Thu, Jun 14, 2012 at 7:59 PM, Igor Stasenko <siguctua at gmail.com> wrote:

>
> On 15 June 2012 03:38, Chris Cunnington <smalltalktelevision at gmail.com>
> wrote:
> >
> >
> > "Any interest?"
> >
> >
> > As ever, David, you are a person to say judicious, reasonable things.
> And what you propose is reasonable, but - alas - it will never happen.
> >
> > I think you're trying to resolve and defuse conflict here, as you
> usually do. And that is praiseworthy, but it is turning away from the
> >
> > politics of this situation.
> >
> Why conflict? I am not conflicting person. I just trying to make my
> point. And i won't kill anyone for it :)
>
> >
> > The players, the ones with actual power, have already decided on an
> immutability bit. This discussion of today is sort of a sideshow.
> >
> > Owing to Andreas's high level of knowledge, it appears, by reading this
> mailing list, that the decision has something to do with him.
> >
> > I think that's like saying if Steph had once offered an opinion on a
> mailing list about how Croquet should have been designed that it would have
> been implemented.
> >
> > Not likely.
> >
>
> That's sad.. So let us wait couple more decades, before someone else
> will do the next step,
> and then crowd will start crying "we want to have the same, we want to
> have the same".
>
> >
> > Bottom line:
> >
> > Steph wants an immutability bit.
> >
> > Eliot has made up his mind. And he won't demo something he already
> implemented in Visual Works.
> >
> > Why is Igor arguing against something his boss wants?
> >
>
> Perhaps because my boss cares to listen before rejecting, and don't
> use a dictator's rule to
> force me doing things, just because he thinks this is right thing to do? :)
>

Whoa, whoa, whoa, whoa, whoa!  Both of you.  I am *not* anyone's boss.  I
don't dictate what happens with the VM.  I did the Cog JIT.  I'm interested
in delivering what I know works much more than I'm interested in
experiments, essentially because that's what I think is most useful to the
community (more useful to have a solid fast VM first, than an experimental,
possibly broken VM sooner).  But I'm not here to stop anyone from doing
anything.  The corollary is that I won't be bullied into doing things I
don't want to do either.  But my /not/ doing something doesn't prevent
someone else from doing something does it?

When I did Cog I was careful not to add lots of new features and to only
remove rarely used features that really got in the way.  I made some
changes (Floats in platform order, one can't change the Character table
Array, but one can change its contents, added closures), but most things
stayed the same and most code worked.  Hence the community now has a
medium-speed VM that's a lot faster than the interpreter and is reliable
and widely used.  In implementing a new object representation I want to do
the same, provide a reliable VM that is even faster, with a few
improvements that I know are useful (immutability, aligning all objects on
an 8-byte boundary for float access through the FFI, lazy become, an
incremental GC, ephemerons, segmented heap growth) and I know can be
implemented.  But my following my agenda (which I think is a useful one) a)
doesn't imply that I think it is the only useful agenda, b) doesn't prevent
people following their own agenda.

So please, each to his own, OK?


>
>
> >
> > Enjoy the immutability bit. It's on its way.
> >
>
> I knew that there's little sense to discuss these things.
> I just thought/had a hope that this discussion is not about
> "let's discuss how to repeat what Eliot did in VW, to make Cog a twin
> brother of it",
> but about a step further from that, because there's always one.
> Yeah, maybe i having too idealistic view on that. Exchanging ideas,
> listen to each other,
> and finally work together to make something real.
> No. That's fallacy. We all should listen to what boss says and do what
> he says. End of story.
>
> Yeah, let us keep making bikes of any color, as long as they black.
>
> Enjoy.
>
> >
> > Chris
> >
>
>
> --
> Best regards,
> Igor Stasenko.
>



-- 
best,
Eliot
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.squeakfoundation.org/pipermail/vm-dev/attachments/20120615/3499c721/attachment.htm