harmony-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Stepan Mishura" <stepan.mish...@gmail.com>
Subject Re: [general] Harmony Q2 release requirements
Date Tue, 06 Mar 2007 05:31:46 GMT
On 3/5/07, Mikhail Loenko wrote:
>
> <SNIP>
> > >>>>> Since currently the most stable platform is Windows/IA32 I
suggest
> > >>>>> that Harmony Q2 will be released on that specific platform
> > >>>>
> > >>>> you mean, we have no time for 2 platforms?
> > >>>
> > >>> I mean we should IMHO make a focus: have superb results on a single
> > >>> platform on a limited set of applications rather than have million
> > >>> somehow working scenarios
> > >>> on a dozen of platforms.
> > >>
> > >> Does it make sense?
> > >
> > > I think, limiting ourselves for the next milestone is a good
> > > idea. Though, IMHO, limiting ourseles to windows is more of a
> > > limitation than of making us focused. It does not take much effort to
> > > support Linux with te same priority of bugfixing (if scenarious are
> > > pretty automated), but lets people be sure that we are not to break
> > > their work in favour to support windows faster.
> > >
> > > There may be a hybrid strategy: improve on windows, do not break
> > > anything on Linux, seems pretty acceptable to me.
> >
> > I have zero interest in exclusively working on Windows.  None.  Zip.
> > Zero.  Nada.
> >
> > Linux is a peer distro for this project.  It always has been.
> >
> > To be an open source project that only distributes software for
> > closed source ecosystems like Windows is the sort of irony I'd prefer
> > not to be associated with :)
>
> Nobody is talking about being a project for Windows only.
>
> The idea is going deep first and extend in deapth next.
>
> If we try to have everything before we have something we will have
> nothing:
> if we try to have presense on each platform before we are solid on at
> least
> one platform we will lose: for each platform there always be "another"
> implementation that is better.
>
> Instead we should stick to some specific platform, make it solid and
> then extend


I don't see a big issue here to ague about - my impression is that most of
the issues currently we have are OS-independent. Why not focus on
architecture rather then OS? We've already implementation for Windows/Linux.
I've heard that Windows implementation a bit more optimized (VM guys please
correct me if I'm wrong). But is it really doesn't matter, IMHO, they both
are quite solid. And I think that improving both of them won't double
required efforts (compared to improving implementation for one selected OS).


Let's say that x86 (Windows/Linux) is our primary focus. And next goal is to
be solid on x86-64 architecture.

Thanks,
Stepan Mishura
Intel Enterprise Solutions Software Division

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message