On Sun, Mar 25, 2007 at 11:43:36PM -0400, Ove Kaaven wrote:
> Robert Millan wrote:
> >OTOH, if we provide win64, game vendors will be very happy that they can
> >target
> >all platforms with a single binary, stablishing win64 as the standard.
>
> Dream on. Game vendors don't care what Wine can and can't run, it's not
> their problem. Their target platforms are it.
And they don't have a target platform for the 64-bits. Sure, 32-bit can live
on, grow extensions and survive for long, but it's fate is doomed. If past the
deadline they're still depending on it, the high end will look for an
alternative. This won't inmediately kill the market, but then it'll be too
late for them: they'll die slowly like 8-bit BASIC or 16-bit MSDOS, trapped in
the past and being incapable to evolve.
> >I suppose we'll have released lenny by end of 2008. In case we haven't
> >(big
> >failure in our part), at that point win64 in wine would stop being a
> >problem and
> >begin being something useful. So my request can be translated into "not
> >providing win64 before end of 2008" rather than "with lenny".
>
> And lenny+1 will be out when? 2009? So if win64 apps appear in 2008,
> debian (stable) won't support them until 2009. Yay.
Not a big deal. The world won't be suddenly populated by win64-only apps
overnight. It'll be a gradual change. Judging from the current rate:
- They haven't started because the ground (a significant portion of 64-bit
capable computers) is not there.
- They haven't even started to prepare the ground (e.g. by releasing 64-bit
Vista).
- They have in fact just released, which means they won't be able to push
for 64-bit computers for a whole release cycle !! And even then it's not
clear they can make it since they lack:
- All the drivers.
- Any interest from users (since there are no win64-only apps). Yes, this
is the chicken&egg problem they already solved for win32, but back then
there was no competition and they could do anything they wanted.
I have my doubts that even lenny+1 users will need win64. But if they do, at
first they'll be a minority. They can find an easy solution by using backports
or getting wine from upstream (which btw, ought to be easier than it is now:
just visit the site and a few clicks).
Anyway, if PAE is going to delay win64 as you suggest below (and I would think
that, too), isn't it a bit unrealistic to want win64 in debian for 2009 ? Your
predicted "hard deadline" is 2014.
> 1) Eric Raymond is so full of crap (again) it's not even funny.
Hehe. Well, we can argue about Eric being full of crap (and I guess we're
pretty much going to agree on that), or we can argue about the content itself.
Sure, Eric might have been using this analisys to push for his particular
agenda (giving up on freedom, etc), but this doesn't mean he isn't making
valid points.
> While
> the whole article is silly, the only thing I even need to mention, is
> that x86-64 systems don't need 64-bit software, they can run 32-bit
> software, and this 32-bit software can address 64GB of RAM (see
> http://en.wikipedia.org/wiki/Physical_Address_Extension for an
> explanation), which, if we even accept his assumptions that RAM usage
> drives OS migration, places the supposed "hard deadline" in 2014.
>
> (MS-DOS, also, could address more than 1MB, and even run 32-bit code,
> such as Doom. Windows took over for quite different reasons.)
Yes, I know about PAE and also remember DOS4GW. Note this part of the article:
"[13] Every platform grows memory extensions as it nears its end of life.
The 8-bit systems had a technique called "bank switching", the DOS machines
limped along with extended/expanded/extruded memory, and these days we have
Intel's Page Addressing Extensions. It doesn't matter; rewriting software
to jump through these hoops is about as much work as porting the software to
a new platform where accessing more memory is natural."
Sure thing, even if its unnatural they can do things like:
- Make kernel use PAE to take advantage of >4GiB, by making each process
run in its own 32-bit space. This is useful in the early post-deadline
era, but won't live for long since big apps need >4GiB themselves.
- Applications *can* use PAE to address >4GiB. This can make the 32-bit
market survive quite longer past the deadline.
This will allow them to keep exploiting the 32-bit market. Maybe even for a
long time (let's assume untill 2014). However, history teaches something here:
"This is the same situation that left Digital Research and CP/M behind during
the 8-bit to 16-bit transition.[17]
[17] Back in the 8-bit world Digital Research owned the dominant disk-based
OS, CP/M. (Systems that didn't have BASIC burned into ROM generally ran it
under CP/M.) Microsoft's DOS began as a clone of CP/M-86 (the 16-bit version
of CP/M) which was then extended with Unix features. During the transition
Digital Research stayed focused on the existing 8-bit market, where it was
making all its money, leaving Microsoft virtually unchallenged in the 16-bit
PC market for the first few years. When the 8-bit market finally collapsed
in 1984, Microsoft and DOS were already entrenched. The remains of Digital
Research were purchased by Novell a few years later."
You can draw a very interesting parallelism from:
1- Digital Research stayed focused on the existing 8-bit market, where it was
making all its money, leaving Microsoft virtually unchallenged in the 16-bit
PC market for the first few years.
2- When the 8-bit market finally collapsed in 1984, Microsoft and DOS were
already entrenched.
to:
1- [in year 2009] Microsoft stayed focused on the existing 32-bit market,
where it was making all its money, leaving GNU/Linux virtually unchallenged
in the 64-bit PC market for the first few years.
2- When the 32-bit market finally collapsed in 2014 (whatsoever), GNU/Linux
and free software were already entrenched.
> 2) If you seriously think every Joe on the street is going to pick up
> Linux just because it supports 64-bit a little bit faster than Windows,
> I've got a nice bridge in San Francisco to sell you.
It's not Joe, it's the vendors. While most 32-bit vendors keep making profit
out of the decaying 32-bit market, small 64-bit vendors [1] will have to pick
whatever that has the most chances to grow so they can grow with it. When
the 64-bit market inevitably grows to the point of extinguishing 32-bit one,
whatever they have chosen will grow with it.
[1] These small 64-bit vendors could be new companies, small divisions in
existing ones, or even the free software community. In the end, it doesn't
matter to this discussion (although I'd love if it were the FS community
;-))
> 3) World domination is not something to toy with. In my opinion, it
> should be deserved by someone/something responsible - and so, if it
> takes surreptitious machinations to achieve it, I don't even want it.
Oh, but wasn't this ment to be a joke? :-) I think of it the same way as
copyleft, just as a parody of Microsoft's monopoly.
--
Robert Millan
My spam trap is [EMAIL PROTECTED] Note: this address is only intended
for spam harvesters. Writing to it will get you added to my black list.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
