Cross-platform development

43 views
Skip to first unread message

Jon Harrop

unread,
Jun 8, 2007, 9:59:47 AM6/8/07
to

Hi!

We recently ported our work from OCaml+Linux to F#+Windows with surprising
ease. We are interested in reproducing this success by diversifying to the
OSX platform.

However, our preliminary study leads us to believe that supporting OSX is
very difficult, perhaps prohibitively difficult for us to use. We would
need to adopt unimpressive languages (like Objective C), revert to previous
programming paradigms (OOP) and reinvent various development tools
ourselves (library bindings).

If we are to embark on a project of this magnitude, we would need a great
deal of assurance that this platform has a future and that there is money
in the Apple market.

A simple Google trend indicates that OSX is now much less popular than a
single Linux distro (Ubuntu):

http://www.google.com/trends?q=osx%2C+ubuntu

However, Apple hardware is extremely expensive, so the few Apple users that
remain must be willing to spend money.

Searching for next-generation GUI developments from Apple indicates that
they are still playing catchup with respect to garbage collection, let
alone a common language run-time or functional reactive GUIs. When will
OSX's GUI technology be brought up to date?

Apple's stock price has continued to rise:

http://www.magiastrology.com/fin_astro/apple_computer.htm

but this has everything to do with the iPod and the iPhone and nothing to do
with OSX:

http://aldoblog.com/images/aapl-2006-01-09.png

The last major development by Apple that I am aware of was the
hardware-accelerated GUI, but this has long since ceased to be impressive.

Most of the development tools for OSX seem to be laggy ports from Linux.

So, would it be worthwhile for us to adopt the OSX platform? Is it circling
the drain or does it have a rosy future?

--
Dr Jon D Harrop, Flying Frog Consultancy
OCaml for Scientists
http://www.ffconsultancy.com/products/ocaml_for_scientists/?usenet

Jolly Roger

unread,
Jun 8, 2007, 10:56:46 AM6/8/07
to
On 2007-06-08 08:59:47 -0500, Jon Harrop <j...@ffconsultancy.com> said:

> We recently ported our work from OCaml+Linux to F#+Windows with surprising
> ease. We are interested in reproducing this success by diversifying to the
> OSX platform.

You're in the wrong news group - you want comp.sys.mac.programmer.help.

--
JR

Dr zara

unread,
Jun 8, 2007, 11:27:18 AM6/8/07
to

"Jon Harrop" <j...@ffconsultancy.com> wrote in message
news:4669624b$0$8747$ed26...@ptn-nntp-reader02.plus.net...

It's a Turd - flush it. And don't listen to that Asshole, Jolly Roger. He
don't know Jack Shit.


nospamatall

unread,
Jun 8, 2007, 12:11:08 PM6/8/07
to

From the way you write, you seem to have made your mind up already.
You're wrong on a lot of the 'facts', so I would suggest further
research of those 'facts'. What you need to know is the installed base
of Macs and how that is increasing. This isn't the kind of info that is
found easily, and it sounds like you've fallen for some FUD.

Your tone implies that you do not like macs. I'd suggest it is not worth
your while, for that reason. No point in working on somrthing you have a
negative attitude about. Either do the research with an open mind or use
the time to improve existing products. If you leave the research to
others it will inevitably contain their bias.

Andy

MuahMan

unread,
Jun 8, 2007, 12:12:33 PM6/8/07
to

"nospamatall" <nospa...@iol.ie> wrote in message
news:f4bv2t$smt$1...@aioe.org...


Don't listen to nospamatall. He is a FUD factory and and F.O.S. In fact
don't listen to any of the Mactards in this group. The Mac is dead, it's
banished from any and all work environments. It's really just a toy computer
for teenage girls and dorm rooms.

Jim

unread,
Jun 8, 2007, 1:30:51 PM6/8/07
to
In article <i-udnYv6GP-gHvTb...@comcast.com>,
"MuahMan" <Mua...@yahoo.com> wrote:

Your moronic post just proves the lies you create, the FUD you try and
spread. Go stop at the vet, they fix things with problems like yours.

Both these links disprove your lies, FUD, and misinformation.

http://www.apple.com/macatwork/

http://www.apple.com/pro/

Jon Harrop

unread,
Jun 8, 2007, 1:43:11 PM6/8/07
to

I already tried there and the response was "we'd rather have no software at
all than cross-platform software".

However, I don't believe that is representative of Mac users. For example,
Wolfram Research sell a significant proportion of Mathematica licenses to
Mac users. Could we do the same?

nospamatall

unread,
Jun 8, 2007, 1:59:58 PM6/8/07
to
MuahMan wrote:

>
> Don't listen to nospamatall. He is a FUD factory and and F.O.S. In fact
> don't listen to any of the Mactards in this group. The Mac is dead, it's
> banished from any and all work environments. It's really just a toy
> computer for teenage girls and dorm rooms.

You've found a friend!

Jolly Roger

unread,
Jun 8, 2007, 2:08:43 PM6/8/07
to
On 2007-06-08 12:43:11 -0500, Jon Harrop <j...@ffconsultancy.com> said:

> Jolly Roger wrote:
>> On 2007-06-08 08:59:47 -0500, Jon Harrop <j...@ffconsultancy.com> said:
>>> We recently ported our work from OCaml+Linux to F#+Windows with
>>> surprising ease. We are interested in reproducing this success by
>>> diversifying to the OSX platform.
>>
>> You're in the wrong news group - you want comp.sys.mac.programmer.help.
>
> I already tried there and the response was "we'd rather have no software at
> all than cross-platform software".

Actually, I just read through your latest posts to
comp.sys.mac.programmer.help, and the response was nothing of the sort.
Actually, most people answered your questions and were helpful to you.
But it seems you largely refused to accept any OOP solution on Mac OS
X, because your current code base is in F# with Windows Forms et al,
where you apparently prefer to have to write code to draw your own user
interface etc - stuff that is light years *behind* what Mac OS X
software development is all about these days. You seem to want F# on
Mac OS X. But Mac OS X developers don't have to draw their own user
interface elements - that's a thing of the past, and we like it that
way. Then you say things like "Mac programmers still seem to be using
C++ and have yet to adopt modern functional languages.", which flies in
the face of reality. Eventually, people there just gave up on you, due
to your apparent hardheadedness and unwillingness to see anything but
what you want to see.

And, interestingly, one response was:

"Guys, Jon Harrop (aka 'we') is a certified troll and spammer.
He is just advertising his business here.
Don't waste your time. Recently he advertised in comp.lang.lisp
and comp.lang.c++."

Oops. It seems you've been exposed.

--
JR

nospamatall

unread,
Jun 8, 2007, 2:10:51 PM6/8/07
to
Jon Harrop wrote:
> Jolly Roger wrote:
>> On 2007-06-08 08:59:47 -0500, Jon Harrop <j...@ffconsultancy.com> said:
>>> We recently ported our work from OCaml+Linux to F#+Windows with
>>> surprising ease. We are interested in reproducing this success by
>>> diversifying to the OSX platform.
>> You're in the wrong news group - you want comp.sys.mac.programmer.help.
>
> I already tried there and the response was "we'd rather have no software at
> all than cross-platform software".
>
> However, I don't believe that is representative of Mac users. For example,
> Wolfram Research sell a significant proportion of Mathematica licenses to
> Mac users. Could we do the same?
>

If there is no competition even a straight port would probably sell. A
good example is Traktor DJ Studio Pro. Using it is like having windows
on your computer, but with none of the actual advantages. It's horribly
laid out, ugly and unintuitive, but it works well and does things no
other piece of software will. That's not to say all Windows software is
like Traktor, often it is just different. Good software fits in with the
OS not just code-wise. Aesthetics and conventions are equally important.

If the software you produce has an equivalent made for Mac OS already,
and that program is mac-like in its implementation, then a lot more work
would be needed to compete with it. As was suggested earlier, a
programmer group might be better, with a different question. Something
along the lines of asking what the experience of others has been in
trying out the mac market. There are probably whole websites devoted to
such things too.

Andy

Timberwoof

unread,
Jun 8, 2007, 2:23:12 PM6/8/07
to
In article <466996a8$0$8744$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> Jolly Roger wrote:
> > On 2007-06-08 08:59:47 -0500, Jon Harrop <j...@ffconsultancy.com> said:
> >> We recently ported our work from OCaml+Linux to F#+Windows with
> >> surprising ease. We are interested in reproducing this success by
> >> diversifying to the OSX platform.
> >
> > You're in the wrong news group - you want comp.sys.mac.programmer.help.
>
> I already tried there and the response was "we'd rather have no software at
> all than cross-platform software".

What an asshole.

> However, I don't believe that is representative of Mac users. For example,
> Wolfram Research sell a significant proportion of Mathematica licenses to
> Mac users. Could we do the same?

Yes, absolutely. I worked at Macromedia in the late '90s. Every product
except one had Mac and Windows versions. The products I worked on,
Director and Shockwave, were specifically designed from the ground up to
be platform-independent. At one time we had four platforms: Mac 68k, Mac
PPC, Widows 16 bit, and Windows 32 bit. Granted, two of them were easily
handled by the compiler, but the nature of the beast required special
tweaks in all of them. The move to OS X required a commitment from
management, but because the cross-platform architecture was already in
place, it was achieved without a lot of drama.

The trick is, as is well-documented, to separate the functions of your
application into UI, system services such as sound and graphics, and the
engine. There are frameworks that do this for you, but they tend to be
big ungainly things with a lowest-common-demonimator approach whose
product is applications that don't feel native on any platform.

Macintosh is a platform where you should pay special attention to UI
requirements. Mac users tend to be discerning and perfectionists. This
article http://www.sonic.net/~mroeder/?q=node/19 is a bit outdated but
it illustrates an important point: it's worth it to make the Mac version
look like a Mac application, not just a sleazy port from Windows or
Linux. (The degree to which this is necessary depends on your audience,
of course. This might not apply as much to a more technical application.)

It is important to hire engineers who are open-minded about all the
platforms. While the engineers I worked with had their preferences, none
of them overtly hated either Mac or Windows. A problem on the Photoshop
team was that the engineers were firmly divided into two camps, Mac and
Windows. They hated each other's platforms and would not share code. It
was difficult for project managers to achieve parity across releases.
Microsoft is big enough that it can afford independent product teams for
Office for Windows and Office for Macintosh. You are probably not.

And then there's how to do QA a shrink-wrapped commercial cross-platform
software product. A lot depends on the basic relationship between Dev
and QA in your organization, but it works best if the QAEs use all the
platforms. That way, you assure equal quality commitment. If one
platform is seen as an unwelcome step-child, then those customers will
abandon it as soon as a committed competitor arrives on the scene.

In any case, success at cross-platform development flows from
management. If you don't have their support and willingness to hire the
right people in Dev and QA, it will never happen.

--
Timberwoof <me at timberwoof dot com> http://www.timberwoof.com

Jon Harrop

unread,
Jun 8, 2007, 2:18:29 PM6/8/07
to
nospamatall wrote:
> If there is no competition even a straight port would probably sell. A
> good example is Traktor DJ Studio Pro. Using it is like having windows
> on your computer, but with none of the actual advantages. It's horribly
> laid out, ugly and unintuitive, but it works well and does things no
> other piece of software will. That's not to say all Windows software is
> like Traktor, often it is just different. Good software fits in with the
> OS not just code-wise. Aesthetics and conventions are equally important.

Absolutely. I agree that this is an excellent idea. However, it is an ideal
that we would like to work towards incrementally. If we see sufficient
sales from OSX software then we would be inclined to put more effort into a
better, native OSX GUI for our software.

> If the software you produce has an equivalent made for Mac OS already,
> and that program is mac-like in its implementation, then a lot more work
> would be needed to compete with it. As was suggested earlier, a
> programmer group might be better, with a different question. Something
> along the lines of asking what the experience of others has been in
> trying out the mac market. There are probably whole websites devoted to
> such things too.

We have a monopoly.

Timberwoof

unread,
Jun 8, 2007, 2:30:28 PM6/8/07
to
In article <466996a8$0$8744$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> Jolly Roger wrote:
> > On 2007-06-08 08:59:47 -0500, Jon Harrop <j...@ffconsultancy.com> said:
> >> We recently ported our work from OCaml+Linux to F#+Windows with
> >> surprising ease. We are interested in reproducing this success by
> >> diversifying to the OSX platform.
> >
> > You're in the wrong news group - you want comp.sys.mac.programmer.help.
>
> I already tried there and the response was "we'd rather have no software at
> all than cross-platform software".
>
> However, I don't believe that is representative of Mac users. For example,
> Wolfram Research sell a significant proportion of Mathematica licenses to
> Mac users. Could we do the same?

What do you actually do? Your web site suggests you publish books.

You'd be interested to know that O'Reilly have moved their entire
publishing system to OS X. They also seem to think that OS X provides a
rich market for their publications.

Craig Koller

unread,
Jun 8, 2007, 2:31:32 PM6/8/07
to
In article <4669624b$0$8747$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> Hi!
>
> We recently ported our work from OCaml+Linux to F#+Windows with surprising
> ease. We are interested in reproducing this success by diversifying to the
> OSX platform.
>
> However, our preliminary study leads us to believe that supporting OSX is
> very difficult, perhaps prohibitively difficult for us to use. We would
> need to adopt unimpressive languages (like Objective C), revert to previous
> programming paradigms (OOP) and reinvent various development tools
> ourselves (library bindings).

I'm not a developer (others can chime in here) but it's surprising that
you would find it easier to move a Linux app to Windows than to
(Unix-based) MacOS.

>
> If we are to embark on a project of this magnitude, we would need a great
> deal of assurance that this platform has a future and that there is money
> in the Apple market.

Sincerely, if you're serious then you should be headed to the WWDC this
weekend rather than asking a Usenet newsgroup.

http://developer.apple.com/wwdc/

>
> A simple Google trend indicates that OSX is now much less popular than a
> single Linux distro (Ubuntu):
>
> http://www.google.com/trends?q=osx%2C+ubuntu

"OSX" is not the best search parameter. Try a few variations (like
"Mac") and you'll see a different outcome.

>
> However, Apple hardware is extremely expensive, so the few Apple users that
> remain must be willing to spend money.

Feature-for-feature, I don't think the facts support this. True, Apple
doesn't offer budget hardware, but they are price competitive in laptops
and desktops for what they offer. And for business users the price
differential is negligible.

If you look at Apple's Server offering, MacOSX Server universal
licensing is much cheaper than MS' offerings, and the XServe has
recently landed in the top ten hardware rankings.

http://www.digitimes.com/news/a20070604PD204.html

>
> Searching for next-generation GUI developments from Apple indicates that
> they are still playing catchup with respect to garbage collection, let
> alone a common language run-time or functional reactive GUIs. When will
> OSX's GUI technology be brought up to date?

WWDC. Do the homework my friend...

>
> Apple's stock price has continued to rise:
>
> http://www.magiastrology.com/fin_astro/apple_computer.htm
>
> but this has everything to do with the iPod and the iPhone and nothing to do
> with OSX:
>
> http://aldoblog.com/images/aapl-2006-01-09.png

Respectfully disagree. Apple sold 1.5 Million Macs in Q1, a record for
the company. Apple Stores are the most successful retail outlets in the
US as relates to sales/square foot. iPods are nice, but they don't pull
the kind of profit the Mac/Macbook/iMac line does.

Market and browser share of Macs are up.

http://www.pcworld.com/article/id,129246/article.html

Since the "Mac/PC" ads have launched, share of the Mac platform has
increased over 40%.

http://www.macnn.com/articles/07/06/08/get.a.mac.wins.award/

[cut]

Again, WWDC...

Jolly Roger

unread,
Jun 8, 2007, 2:33:48 PM6/8/07
to
On 2007-06-08 13:18:29 -0500, Jon Harrop <j...@ffconsultancy.com> said:

> Absolutely. I agree that this is an excellent idea. However, it is an ideal
> that we would like to work towards incrementally. If we see sufficient
> sales from OSX software then we would be inclined to put more effort into a
> better, native OSX GUI for our software.

You won't *get* good sales if the application doesn't have a decent
user interface. Mac users, in general, don't settle for crap user
interfaces.

--
JR

Jon Harrop

unread,
Jun 8, 2007, 2:28:00 PM6/8/07
to
MuahMan wrote:
> The Mac is dead...

Is this really true?

What new software came out for the Mac this year?

What was the last major development announced by Apple?

Jon Harrop

unread,
Jun 8, 2007, 2:31:29 PM6/8/07
to
Jolly Roger wrote:
> On 2007-06-08 12:43:11 -0500, Jon Harrop <j...@ffconsultancy.com> said:
>> I already tried there and the response was "we'd rather have no software
>> at all than cross-platform software".
>
> Actually, I just read through your latest posts to
> comp.sys.mac.programmer.help, and the response was nothing of the sort.
> Actually, most people answered your questions and were helpful to you.
> But it seems you largely refused to accept any OOP solution on Mac OS
> X, because your current code base is in F# with Windows Forms et al,

Yes.

> where you apparently prefer to have to write code to draw your own user
> interface etc - stuff that is light years *behind* what Mac OS X
> software development is all about these days. You seem to want F# on
> Mac OS X. But Mac OS X developers don't have to draw their own user
> interface elements - that's a thing of the past,

Not in scientific visualization.

> and we like it that
> way. Then you say things like "Mac programmers still seem to be using
> C++ and have yet to adopt modern functional languages.", which flies in
> the face of reality.

There are no OSX native OCaml bindings to anything useful, AFAICT. The only
usable developer tools that exist in this area are ports of GL- and
GTK-related stuff lifted from Linux. Looks like a GTK-based GUI compiled on
a Mac is our best bet then.

I think it would really help if Apple make some language-agnostic developer
tools to help with cross-platform GUI development.

> Eventually, people there just gave up on you, due
> to your apparent hardheadedness and unwillingness to see anything but
> what you want to see.

I consider "you need to ditch your existing work and rewrite everything in
Perl or Objective C" as unhelpful. We're obviously not going to adopt a
sucky language just because that is all OSX offers.

Timberwoof

unread,
Jun 8, 2007, 2:43:15 PM6/8/07
to
In article <cwkollertwo-D18C...@news.giganews.com>,
Craig Koller <cwkol...@netscape.net> wrote:

> In article <4669624b$0$8747$ed26...@ptn-nntp-reader02.plus.net>,
> Jon Harrop <j...@ffconsultancy.com> wrote:
>
> > Hi!
> >
> > We recently ported our work from OCaml+Linux to F#+Windows with surprising
> > ease. We are interested in reproducing this success by diversifying to the
> > OSX platform.
> >
> > However, our preliminary study leads us to believe that supporting OSX is
> > very difficult, perhaps prohibitively difficult for us to use. We would
> > need to adopt unimpressive languages (like Objective C), revert to previous
> > programming paradigms (OOP) and reinvent various development tools
> > ourselves (library bindings).
>
> I'm not a developer (others can chime in here) but it's surprising that
> you would find it easier to move a Linux app to Windows than to
> (Unix-based) MacOS.

More than anything else, it has to do with how similar the UI model is.
Windows 16 was actually pretty similar to Mac OS; I helped develop
Pascal code that would, with proper variable structures and ifdefs,
compile and run on either platform.

OS X UI is object-oriented using Objective C. The code for that could be
quite different from what is needed for Windows or X programming.


<snip>

> Sincerely, if you're serious then you should be headed to the WWDC this
> weekend rather than asking a Usenet newsgroup.
>
> http://developer.apple.com/wwdc/

What he said.

Jolly Roger

unread,
Jun 8, 2007, 2:44:29 PM6/8/07
to
On 2007-06-08 13:31:29 -0500, Jon Harrop <j...@ffconsultancy.com> said:

> Jolly Roger wrote:
>> Eventually, people there just gave up on you, due
>> to your apparent hardheadedness and unwillingness to see anything but
>> what you want to see.
>
> I consider "you need to ditch your existing work and rewrite everything in
> Perl or Objective C" as unhelpful. We're obviously not going to adopt a
> sucky language just because that is all OSX offers.

And, likewise, Mac users are obviously not going to want your crappy
software just because you don't feel like trying. Bye bye now.

--
JR

Jolly Roger

unread,
Jun 8, 2007, 2:46:45 PM6/8/07
to
On 2007-06-08 13:31:32 -0500, Craig Koller <cwkol...@netscape.net> said:

> Respectfully disagree. Apple sold 1.5 Million Macs in Q1, a record for
> the company. Apple Stores are the most successful retail outlets in the
> US as relates to sales/square foot. iPods are nice, but they don't pull
> the kind of profit the Mac/Macbook/iMac line does.
>
> Market and browser share of Macs are up.
>
> http://www.pcworld.com/article/id,129246/article.html
>
> Since the "Mac/PC" ads have launched, share of the Mac platform has
> increased over 40%.
>
> http://www.macnn.com/articles/07/06/08/get.a.mac.wins.award/

He doesn't care. He's got his mind already made up. And frankly, Mac
users won't settle for the only kind of crap ware he is willing to
write anyway. Good riddance if you ask me.

--
JR

MuahMan

unread,
Jun 8, 2007, 2:47:36 PM6/8/07
to

"Jon Harrop" <j...@ffconsultancy.com> wrote in message
news:4669a129$0$8744$ed26...@ptn-nntp-reader02.plus.net...


The last major development announcement made by Apple was to delay Leopard
and put it on the back burner while they further developed the gadget market
for iPods and AppleTV.

nospamatall

unread,
Jun 8, 2007, 2:49:53 PM6/8/07
to
Yes, generally, but that was the point I made about Traktor. It sucks,
interface-wise, isn't even a particularly good windows program, but it's
all there is that does what it does, so it sells.

All it will take though, is for someone to come along with a half-decent
interface to an equally functional program and they will be gone.
Obviously hasn't been worth their while so far, but as mac market share
increases that might change. I can see how it would be a lot of work.

Andy

Timberwoof

unread,
Jun 8, 2007, 2:50:41 PM6/8/07
to
In article <46699eed$0$8735$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> nospamatall wrote:
> > If there is no competition even a straight port would probably sell. A
> > good example is Traktor DJ Studio Pro. Using it is like having windows
> > on your computer, but with none of the actual advantages. It's horribly
> > laid out, ugly and unintuitive, but it works well and does things no
> > other piece of software will. That's not to say all Windows software is
> > like Traktor, often it is just different. Good software fits in with the
> > OS not just code-wise. Aesthetics and conventions are equally important.
>
> Absolutely. I agree that this is an excellent idea. However, it is an ideal
> that we would like to work towards incrementally. If we see sufficient
> sales from OSX software then we would be inclined to put more effort into a
> better, native OSX GUI for our software.

This is a good formula to follow if you don't really want to commit to
the project, but want to demonstrate to management that it won't make
money. It works like this: if your first release in the Mac market is
too Windows-like, then it will not sell well. You make minor
improvements to the Mac version, mostly added features to keep up with
the Windows version. Don't market it because you don't have many
customers, and the release will sell even fewer copies. Eventually, it
will not meet any of the criteria to keep it alive, and you can cancel
it.

> > If the software you produce has an equivalent made for Mac OS already,
> > and that program is mac-like in its implementation, then a lot more work
> > would be needed to compete with it. As was suggested earlier, a
> > programmer group might be better, with a different question. Something
> > along the lines of asking what the experience of others has been in
> > trying out the mac market. There are probably whole websites devoted to
> > such things too.
>
> We have a monopoly.

If you think that you can hold that monopoly while releasing substandard
software, you're mistaken. Some bright programmer will see your product,
realize he can do it better and more Mac-like, and take your market from
you.

Please excuse my tone. I have worked for several companies that tried,
with varying success, to bring Mac products to market. I know many of
the strategies and I know what works. I'm not inclined to sugar-coat the
information for a disinterested management team. If you don't really
care, then you might as well not waste your time with it.

Jon Harrop

unread,
Jun 8, 2007, 2:53:57 PM6/8/07
to
nospamatall wrote:

> Jolly Roger wrote:
>> You won't *get* good sales if the application doesn't have a decent user
>> interface. Mac users, in general, don't settle for crap user interfaces.
>>
> Yes, generally, but that was the point I made about Traktor. It sucks,
> interface-wise, isn't even a particularly good windows program, but it's
> all there is that does what it does, so it sells.

Exactly. I think this is more representative of the kind of work we do
(scientific computing and scientific visualization). Programs like
Mathematica are an excellent case study here.

> All it will take though, is for someone to come along with a half-decent
> interface to an equally functional program and they will be gone.
> Obviously hasn't been worth their while so far, but as mac market share
> increases that might change. I can see how it would be a lot of work.

I'm willing to risk this.

Craig Koller

unread,
Jun 8, 2007, 3:07:13 PM6/8/07
to
In article
<timberwoof.spam-A0...@nnrp-virt.nntp.sonic.net>,
Timberwoof <timberw...@inferNOnoSPAMsoft.com> wrote:

> In article <cwkollertwo-D18C...@news.giganews.com>,
> Craig Koller <cwkol...@netscape.net> wrote:
>
> > In article <4669624b$0$8747$ed26...@ptn-nntp-reader02.plus.net>,
> > Jon Harrop <j...@ffconsultancy.com> wrote:
> >
> > > Hi!
> > >
> > > We recently ported our work from OCaml+Linux to F#+Windows with
> > > surprising
> > > ease. We are interested in reproducing this success by diversifying to
> > > the
> > > OSX platform.
> > >
> > > However, our preliminary study leads us to believe that supporting OSX is
> > > very difficult, perhaps prohibitively difficult for us to use. We would
> > > need to adopt unimpressive languages (like Objective C), revert to
> > > previous
> > > programming paradigms (OOP) and reinvent various development tools
> > > ourselves (library bindings).
> >
> > I'm not a developer (others can chime in here) but it's surprising that
> > you would find it easier to move a Linux app to Windows than to
> > (Unix-based) MacOS.
>
> More than anything else, it has to do with how similar the UI model is.
> Windows 16 was actually pretty similar to Mac OS; I helped develop
> Pascal code that would, with proper variable structures and ifdefs,
> compile and run on either platform.

Yes, Macromedia's (Director's) cross-platform capabilities enabled me to
move from a Mac-only shop in the mid-90s to the Mac-dev/Win-deploy job
I've been at for the past 12 years. I don't know whether to thank you or
blame you. ;-)

Steve Hix

unread,
Jun 8, 2007, 3:15:25 PM6/8/07
to
In article <4669a129$0$8744$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> MuahMan wrote:
> > The Mac is dead...
>
> Is this really true?

Just in case you hadn't noticed, "Muahman" (Brian Pratt) is a long-time
and particularly noxious troll in this groupl



> What new software came out for the Mac this year?

Gargle is your friend. (You can also check Apple's website pages
covering software for the platform.



> What was the last major development announced by Apple?

Check Apple's website.

Jon Harrop

unread,
Jun 8, 2007, 3:37:50 PM6/8/07
to
Craig Koller wrote:
> I'm not a developer (others can chime in here) but it's surprising that
> you would find it easier to move a Linux app to Windows than to
> (Unix-based) MacOS.

I was pretty amazed myself. :-)

The code is entirely OCaml (250kLOC) using OpenGL. It compiles and runs
flawlessly on PPC and Intel Macs but lacks the rest of the GUI, which was
<1% of the work on Linux and Windows but we don't even know where to start
with OSX.

>> If we are to embark on a project of this magnitude, we would need a great
>> deal of assurance that this platform has a future and that there is money
>> in the Apple market.
>
> Sincerely, if you're serious then you should be headed to the WWDC this
> weekend rather than asking a Usenet newsgroup.
>
> http://developer.apple.com/wwdc/

I wouldn't mind doing that if I had more faith that our software had a
future on OSX. This project is really just me putting feelers out to see if
there is any market there. I am very glad to hear a positive response.

>> A simple Google trend indicates that OSX is now much less popular than a
>> single Linux distro (Ubuntu):
>>
>> http://www.google.com/trends?q=osx%2C+ubuntu
>
> "OSX" is not the best search parameter. Try a few variations (like
> "Mac") and you'll see a different outcome.

Ah, ok. Right, that looks much more promising. :-)

http://www.google.com/trends?q=mac%2C+linux

>> However, Apple hardware is extremely expensive, so the few Apple users
>> that remain must be willing to spend money.
>
> Feature-for-feature, I don't think the facts support this. True, Apple
> doesn't offer budget hardware,

I think the Mac Mini is impressive.

> If you look at Apple's Server offering, MacOSX Server universal
> licensing is much cheaper than MS' offerings, and the XServe has
> recently landed in the top ten hardware rankings.
>
> http://www.digitimes.com/news/a20070604PD204.html

I compared hardware with my Linux box. Apparently £3,000 will get me
something slightly worse in every respect than last year's £1,600 Linux
box.

I haven't bothered looking into Windows servers. Frankly I think it'd be
0wn3d in <3 mins were it not behind my Linux firewall. ;-)

>> Apple's stock price has continued to rise:
>>
>> http://www.magiastrology.com/fin_astro/apple_computer.htm
>>
>> but this has everything to do with the iPod and the iPhone and nothing to
>> do with OSX:
>>
>> http://aldoblog.com/images/aapl-2006-01-09.png
>
> Respectfully disagree. Apple sold 1.5 Million Macs in Q1,

Wow. Ok, that's incredible. I'll keep looking at Apple. :-)

ZnU

unread,
Jun 8, 2007, 4:40:42 PM6/8/07
to
In article <4669a73e$0$8743$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> nospamatall wrote:
> > Jolly Roger wrote:
> >> You won't *get* good sales if the application doesn't have a decent user
> >> interface. Mac users, in general, don't settle for crap user interfaces.
> >>
> > Yes, generally, but that was the point I made about Traktor. It sucks,
> > interface-wise, isn't even a particularly good windows program, but it's
> > all there is that does what it does, so it sells.
>
> Exactly. I think this is more representative of the kind of work we do
> (scientific computing and scientific visualization). Programs like
> Mathematica are an excellent case study here.
>
> > All it will take though, is for someone to come along with a half-decent
> > interface to an equally functional program and they will be gone.
> > Obviously hasn't been worth their while so far, but as mac market share
> > increases that might change. I can see how it would be a lot of work.
>
> I'm willing to risk this.

If you're just looking for a technical solution to get your UI up and
running, and have no particular interest in making your app Mac-like,
why not use the same libraries you're using on Linux and give Mac users
the same X11-based interface you're presumably delivering there?

--
"That's George Washington, the first president, of course. The interesting thing
about him is that I read three--three or four books about him last year. Isn't
that interesting?"
- George W. Bush to reporter Kai Diekmann, May 5, 2006

Craig Koller

unread,
Jun 8, 2007, 5:17:57 PM6/8/07
to
In article <4669b187$0$8745$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:
[cut]

>
> I compared hardware with my Linux box. Apparently £3,000 will get me
> something slightly worse in every respect than last year's £1,600 Linux
> box.
>
Can't speak to Apple's price competitiveness outside of the US, given
all the other trade/currency variables. Also, considering that they
compete with Dell, HP, Sony and the other majors, Apple is normally
compared (and is seen favorably) from the name brand perspective.

Ironically, I left this newsgroup and found this article shortly
thereafter:

http://www.computerworld.com/action/article.do?command=viewArticleBasic&a
rticleId=9023959

Worth a read.

Mitch

unread,
Jun 8, 2007, 6:22:22 PM6/8/07
to
In article <4669a1fa$0$8744$ed26...@ptn-nntp-reader02.plus.net>, Jon
Harrop <j...@ffconsultancy.com> wrote:

> But Mac OS X developers don't have to draw their own user
> > interface elements - that's a thing of the past,
>
> Not in scientific visualization.

This is a weird comment.
You want to build every part of your own interface elements because
you need control over what is somewhere else?
Apple wants you to build with their UI elements because that is what
the UI is based on -- without them, your product will look like junk
and work less well.
And if you're deciding anything based on how the UI will look when you
finish, you're looking at the wrong end of the project.

Jon Harrop

unread,
Jun 8, 2007, 7:11:01 PM6/8/07
to
ZnU wrote:
> If you're just looking for a technical solution to get your UI up and
> running, and have no particular interest in making your app Mac-like,
> why not use the same libraries you're using on Linux and give Mac users
> the same X11-based interface you're presumably delivering there?

Yes, I think that is probably our best bet. The GTK bindings for OCaml have
already been ported to Mac but there are no bindings to either Cocoa or
Carbon. Assuming GLArea is supported by the Mac GTK bindings, that should
be enough.

Jon Harrop

unread,
Jun 8, 2007, 7:48:57 PM6/8/07
to
Craig Koller wrote:
> Can't speak to Apple's price competitiveness outside of the US, given
> all the other trade/currency variables. Also, considering that they
> compete with Dell, HP, Sony and the other majors, Apple is normally
> compared (and is seen favorably) from the name brand perspective.
>
> Ironically, I left this newsgroup and found this article shortly
> thereafter:
>
> http://www.computerworld.com/action/article.do?command=viewArticleBasic&a
> rticleId=9023959
>
> Worth a read.

This seems to be a geographical thing. I just compared prices in detail and
your absolutely right with region set to US. The story is very different
here in the UK though.

The cheapest Mac Pro is:

£1,699
2x 2x 2GHz Xeon
1GB RAM
nVidia GF7300GT 256Mb
250Gb HD

A similar Dell gives:

£1,699
4x 2.4GHz Q6600
2Gb RAM
nVidia GF8800GTX 768Mb
500Gb RAM

That's a faster CPU, twice the RAM, twice the HD and three times the video
memory of the Apple Mac for the same price!

John C. Randolph

unread,
Jun 8, 2007, 8:57:34 PM6/8/07
to
On 06/08/07, Jolly Roger <jolly...@R.E.M.O.V.E.pobox.com> said:

> On 2007-06-08 08:59:47 -0500, Jon Harrop <j...@ffconsultancy.com> said:
>
>> We recently ported our work from OCaml+Linux to F#+Windows with surprising
>> ease. We are interested in reproducing this success by diversifying to the
>> OSX platform.
>

> You're in the wrong news group - you want comp.sys.mac.programmer.help.

No, he doesn't want help. He's trolling.

-jcr

John C. Randolph

unread,
Jun 8, 2007, 8:57:44 PM6/8/07
to
On 06/08/07, Jon Harrop <j...@ffconsultancy.com> said:

> We would
> need to adopt unimpressive languages (like Objective C), revert to previous
> programming paradigms (OOP) and reinvent various development tools
> ourselves (library bindings).

You should not port your product to Mac OS X.

Your contempt for your potential customers on the Mac is palpable, and
it will be reflected in your efforts: you will depart from the
standards of the platform, you will not bother to make your application
accessible, scriptable, or interoperable with other applications, you
will complain on developer lists about how you run into brick walls
when refusing to learn how to use the native tools properly, and you
will generally make a nuisance of yourself.

Ports in general are half-assed attempts to generate incremental
revenue, and are typically supported (if you want to even call it that)
by vendors who know little to nothing about the platform.

If you have users demanding your product on OS X, then find a company
that's quite familiar with developing native OS X applications, and
hire them to create a program that is compatible with your data format.

-jcr

Tim Adams

unread,
Jun 8, 2007, 9:32:23 PM6/8/07
to
In article <4669ec61$0$8732$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

Are you sure the Q6600, a core 2 duo chip, is faster then the Xeon?

> twice the RAM, twice the HD and three times the video
> memory of the Apple Mac for the same price!

--
regarding Snit "You are not flamed because you speak the truth,
you are flamed because you are a hideous troll and keep disrupting
the newsgroup." Andrew J. Brehm

gimme_this...@yahoo.com

unread,
Jun 8, 2007, 11:11:53 PM6/8/07
to
You're approaching the issue from the point of view of someone who
thinks OS X is a dialect of Linux.

Background:

Windows software sells because users know that once they get results
they can post those results into Word, Excel, Pdf, LaTex, Scientific
Word, Mathematica, whatever is applicable for that area of scientific
research.

Consequently:

If you build for OS X you should research what's available on OS X
that make it desirable to develop in. That is, what does OS X have the
Linux and Windows do not have. The answer is that Apple has
Applescript. You can think of AppleScript as the equivalent of VBA for
Office, but the difference is that many venders, not just one, support
it. iLife application support it, EyeTV supports it. There is a
secondary answer as well. OS X has Automator.

So you should ask, if we develop for OS X, what will users be able to
do on that platform that they can not do on any other?

Is the appeal of what it can do make it worth the trouble of porting?

If not. Don't do it. Mac users can use Parallels and run your app on
either Ubunto or XP.

Some of what you say is totally off the wall.

If you're geniunely concerned that Apple hardware is to expensive,
then get real. Any enterprise can afford to provide its workers a
$1000 laptop, or even a $2000 desktop. If you're users can't afford
$1000 for their machine, it's unlikely they are going to spend more
then $40 for your product.

Anyway, companies often spend $30,000 for a database license or
$50,000 for a WebLogic license. The point is, if your software does
something unique and people can easily use it, they'll be willing to
spend a lot of money for it.

You also added some Troll assertions about Mac OS X GUI software. The
basic API was significantly improved when Apple migrated to UNIX in
the early 2000s. If you look at Expose, at you can see files on the
dock, when you never see anything like a Windows Mirror Effect, it's
obvious even to non technical people that Mac's GUI is state of the
art.

Timberwoof

unread,
Jun 9, 2007, 1:37:04 AM6/9/07
to
In article <cwkollertwo-65FD...@news.giganews.com>,
Craig Koller <cwkol...@netscape.net> wrote:

> > More than anything else, it has to do with how similar the UI model is.
> > Windows 16 was actually pretty similar to Mac OS; I helped develop
> > Pascal code that would, with proper variable structures and ifdefs,
> > compile and run on either platform.
>
> Yes, Macromedia's (Director's) cross-platform capabilities enabled me to
> move from a Mac-only shop in the mid-90s to the Mac-dev/Win-deploy job
> I've been at for the past 12 years. I don't know whether to thank you or
> blame you. ;-)

I didn't write it; I tested it. :-)

Timberwoof

unread,
Jun 9, 2007, 1:41:49 AM6/9/07
to
In article <f4c8ci$r1a$1...@aioe.org>, nospamatall <nospa...@iol.ie>
wrote:

It is an error to consider only Mac market share. You have to also
consider the number of competitors you have in each market. For
instance, in Windows, if you want to make some software that Microsoft
already makes, people will think you're nuts. But if you make the same
software for OS X, say a browser or presentation software, then you can
sell many copies. Would you rather be a medium-sized fish in a big pond
or the same size fish in a small pond?

Craig Koller

unread,
Jun 9, 2007, 2:33:41 AM6/9/07
to
In article
<teadams$2$0$0$3-4C55A6.21...@news.west.earthlink.net>,
Tim Adams <teadams$2$0$0$3...@earthlink.net> wrote:

That 250GB of HD space is $50.
That GB of RAM is $80
That 512 MB of Video RAM is worth what to a non-gamer?

You're talking $200 - $300 max. And you don't get the iLife apps,
probably don't get the digital audio outs, FW800, etc. And you don't get
the build quality of the Macs, aluminum tanks complete with four
tray-loading SATA bays and 32GB RAM capacity in the easiest-to-upgrade
form factor I've experienced. And finally there's Apple support, where
Dell doesn't come close:

http://www.linuxelectrons.com/news/hardware/apple-ranks-highest-free-tech
-support-says-survey

Again, feature-for-feature, the idea that Macs are more expensive is a
myth. In my experience you pay more for the PC, either up front or in
lost productivity and troubleshooting.

Seriously, these comparisons have been done ad nauseam over the years.
For the record, Apple has fewer models and incremental upgrades than
most, so usually when a model is reaching the end of its cycle is where
the others catch or pass it slightly on price, but then as soon as the
new models come out, the meters are reset with Apple offering the best
value and it starts all over again. The new MacBooks have just been
refreshed. It's been about a year since the current Mac Pros have been
out. I imagine we'll be hearing about an upgrade very soon.

Seriously Jon - let the price thing go. It's a transparent and frankly
an amateurish argument. If your users are serious and serve a valid
business/scientific need, they'll easily generate enough value to cover
$2-300 on the first engagement...

Jon Harrop

unread,
Jun 9, 2007, 8:08:46 AM6/9/07
to
Craig Koller wrote:
> That 250GB of HD space is $50.

Apple charge £79, which is almost $160.

> That GB of RAM is $80

Apple charge £189, which is almost $380.

> That 512 MB of Video RAM is worth what to a non-gamer?

Everything.

Apple's graphics hardware has really shocked me actually. The best graphics
hardware they offer in the Mac Pro couldn't run the stuff I've been working
on for the past 6 months. Aside from the small video RAM, the GPU itself is
a generation out of date. Worse than that, it isn't even a good 7 series.

Just to put this into perspective, the machine I'm working on cost £800 a
year ago and it outspecs that £1,699 Mac in every department except CPU. In
particular, it has a 7900 with 512Mb RAM.

> Seriously Jon - let the price thing go. It's a transparent and frankly
> an amateurish argument.

On the contrary, I think it is an objective and quantitative comparison.
Even if you compare a Mac to a prebuilt PC like a Dell in the UK, I just
cannot see how anyone can claim that Mac hardware doesn't cost more.

Is it possible to order a US Mac and have it delivered to the UK? That could
knock $900 off for a start...

Also, am I correct to assume that I can upgrade the RAM and HD of a Mac but
not the graphics card?

Jon Harrop

unread,
Jun 9, 2007, 8:19:14 AM6/9/07
to
gimme_this...@yahoo.com wrote:
> So you should ask, if we develop for OS X, what will users be able to
> do on that platform that they can not do on any other?

That is a nice ideal but I'd like to get there incrementally.

> If not. Don't do it. Mac users can use Parallels and run your app on
> either Ubunto or XP.

Does it support hardware accelerated graphics? For example, can you run
Windows games on the Mac?

> If you're geniunely concerned that Apple hardware is to expensive,

I objected to the claim that Apple hardware is as cheap as a PC.

> You also added some Troll assertions about Mac OS X GUI software. The
> basic API was significantly improved when Apple migrated to UNIX in
> the early 2000s. If you look at Expose, at you can see files on the
> dock, when you never see anything like a Windows Mirror Effect, it's
> obvious even to non technical people that Mac's GUI is state of the
> art.

I'm sure Mac software is amazing to look at. I was referring to the lack of
developer tools.

Tim Adams

unread,
Jun 9, 2007, 9:56:38 AM6/9/07
to
In article <466a99c6$0$8733$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

~snip

>
> Also, am I correct to assume that I can upgrade the RAM and HD of a Mac but
> not the graphics card?

No, you're not safe to assume that. depending on the Mac you buy, you can
upgrade all three of the items you list. In fact, with the Mac Pro you can have
multiple graphics cards driving multiple displays.

ZnU

unread,
Jun 9, 2007, 11:16:43 AM6/9/07
to
In article <466a9c39$0$8722$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> gimme_this...@yahoo.com wrote:
> > So you should ask, if we develop for OS X, what will users be able to
> > do on that platform that they can not do on any other?
>
> That is a nice ideal but I'd like to get there incrementally.
>
> > If not. Don't do it. Mac users can use Parallels and run your app on
> > either Ubunto or XP.
>
> Does it support hardware accelerated graphics? For example, can you run
> Windows games on the Mac?

This was added in 3.0, which just came out a couple of days ago:
http://www.parallels.com/en/products/desktop/

[snip]

Jolly Roger

unread,
Jun 9, 2007, 12:01:56 PM6/9/07
to
On 2007-06-09 07:19:14 -0500, Jon Harrop <j...@ffconsultancy.com> said:

>> If not. Don't do it. Mac users can use Parallels and run your app on
>> either Ubunto or XP.
>
> Does it support hardware accelerated graphics? For example, can you run
> Windows games on the Mac?

Can you not read?:

<http://www.parallels.com/en/products/desktop/>

--
JR

Jon Harrop

unread,
Jun 9, 2007, 1:12:54 PM6/9/07
to
ZnU wrote:
> This was added in 3.0, which just came out a couple of days ago:
> http://www.parallels.com/en/products/desktop/

Great. Any indication on how reliable it is or what shortcomings it has? In
particular, how much of the video memory is left for the emulated program?

Jon Harrop

unread,
Jun 9, 2007, 1:34:45 PM6/9/07
to
Tim Adams wrote:
> No, you're not safe to assume that. depending on the Mac you buy, you can
> upgrade all three of the items you list.

If I buy a Mac, can I upgrade the graphics card to an 8-series GeForce with
>=512Mb RAM? I believe not, but I may be wrong.

MuahMan

unread,
Jun 9, 2007, 1:51:18 PM6/9/07
to

"Jolly Roger" <jolly...@R.E.M.O.V.E.pobox.com> wrote in message
news:2007060813442950878-jollyroger@REMOVEpoboxcom...
> On 2007-06-08 13:31:29 -0500, Jon Harrop <j...@ffconsultancy.com> said:
>
>> Jolly Roger wrote:
>>> Eventually, people there just gave up on you, due
>>> to your apparent hardheadedness and unwillingness to see anything but
>>> what you want to see.
>>
>> I consider "you need to ditch your existing work and rewrite everything
>> in
>> Perl or Objective C" as unhelpful. We're obviously not going to adopt a
>> sucky language just because that is all OSX offers.
>
> And, likewise, Mac users are obviously not going to want your crappy
> software just because you don't feel like trying. Bye bye now.
>
> --
> JR
>

LOL, Jolly Roger is pissed because OS X has horrible development tools. That
and she has sand in her Vagina.

Timberwoof

unread,
Jun 9, 2007, 1:58:03 PM6/9/07
to
In article <466ae62b$0$8711$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> Tim Adams wrote:
> > No, you're not safe to assume that. depending on the Mac you buy, you can
> > upgrade all three of the items you list.
>
> If I buy a Mac, can I upgrade the graphics card to an 8-series GeForce with
> >=512Mb RAM? I believe not, but I may be wrong.

it depends on what kind you get. You can replace the video card in any
G4 or G5 tower. Whether it's the specific card you listed, I don't know
offhand, but there are plenty of high-zoot video cards available. Check
xlr8yourmac.com for compatibility databases.

Timberwoof

unread,
Jun 9, 2007, 2:01:12 PM6/9/07
to
In article <466ae10d$0$8730$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> ZnU wrote:
> > This was added in 3.0, which just came out a couple of days ago:
> > http://www.parallels.com/en/products/desktop/
>
> Great. Any indication on how reliable it is or what shortcomings it has? In
> particular, how much of the video memory is left for the emulated program?

I don't understand the question. That is, I don't understand the
assumptions you're making in formulating the question.

Parallels runs in a few modes, the most exciting of which is that
Windows applications run in the full Windows environment, but their
windows and UI appear entirely within the OS X desktop. The CPU to run
the Windows processes is not emulated; the software runs directly on the
Mac's own CPU. However, it uses main RAM to run, not video RAM. That's
generally reserved for things like video (and, in Windows, some printing
functions).

Tim Adams

unread,
Jun 9, 2007, 4:17:09 PM6/9/07
to
In article <466ae62b$0$8711$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> Tim Adams wrote:
> > No, you're not safe to assume that. depending on the Mac you buy, you can
> > upgrade all three of the items you list.
>
> If I buy a Mac, can I upgrade the graphics card to an 8-series GeForce with
> >=512Mb RAM? I believe not, but I may be wrong.

The Mac Pro should support it but you would need to check the video card
company for a driver.

Jon Harrop

unread,
Jun 9, 2007, 4:41:02 PM6/9/07
to
Timberwoof wrote:
> Parallels runs in a few modes, the most exciting of which is that
> Windows applications run in the full Windows environment, but their
> windows and UI appear entirely within the OS X desktop. The CPU to run
> the Windows processes is not emulated; the software runs directly on the
> Mac's own CPU. However, it uses main RAM to run, not video RAM. That's
> generally reserved for things like video (and, in Windows, some printing
> functions).

Yes, fast-tracking a virtualized CPU is the easy part. Nowadays you can do
it in hardware using Intel VT.

Virtualizing the GPU is much harder and all other attempts that I am aware
of have failed (except an alpha-release of Chromium for Xen).

I have heard that Parallels does a great job on CPU virtualization, and I
can well believe that because it is now a tried and tested technique.
However, GPU virtualization is very new and very difficult. I believe
Parallels are doing pioneering work here and I'd like to know how well
their latest implementation works.

You are correct that I did make several assumptions about their
implementation. I assumed that it is not possible to do any virtualization
on the GPU, so the task of virtualizing is moved either into the driver or
into an abstraction layer behind the driver. Depending how this is done,
the cost of that abstraction could be crippling, e.g. emulating PCI-E
transfers. There is also the cost of sharing video memory between virtual
hosts. Finally, high-performance video drivers are a hugely difficult
software engineering challenge and require great sophistication. If
Parallels somehow licensed nVidia's drivers and butchered them to get the
best performance then they will almost certainly have introduced bugs.

I would not speculate as to how this is actually implemented or even what
the performance trade-offs would be. However, I would like to see any hard
numbers on how fast Parallels current implementation runs, particularly
under high video memory usage (e.g. Doom 3 in Ultra quality on a 512Mb
card).

There are various articles out there but nothing I would believe:

"Parallels doesn't officially support DirectX yet, but that didn't stop us
from trying to run Quake 4 on it anyway. Alas, it was an exercise in
futility--it did not work--and in fact, at one point, even forced a
spontaneous system reboot!" -
http://reviews.cnet.com/4531-10921_7-6546370.html

This article doesn't seem to be very accurate but I can well believe that a
virtualized GPU undermines the stability of all virtual hosts running on
that physical host.

Jon Harrop

unread,
Jun 9, 2007, 5:09:30 PM6/9/07
to
Tim Adams wrote:
> The Mac Pro should support it but you would need to check the video card
> company for a driver.

Just looking at nVidia's website, they seem to have drivers for Windows,
Linux, Solaris and even FreeBSD but not Mac:

http://www.nvidia.com/content/drivers/drivers.asp

There are nVidia driver updates bundled with Apple's OSX updates:

http://docs.info.apple.com/article.html?artnum=120288

but they do not mention modern graphics cards.

Here's a thread about a guy who can only get an 8800 to work in his Mac Pro
by running Windows instead of OSX:

http://forums.macrumors.com/showthread.php?t=250851

I am very surprised that nVidia support Solaris and FreeBSD but not Mac.
Given the Mac's legacy, I'd have thought they'd put more effort into
keeping up to date with respect to graphics hardware.

George Graves

unread,
Jun 9, 2007, 5:57:38 PM6/9/07
to
On Sat, 9 Jun 2007 10:58:03 -0700, Timberwoof wrote
(in article
<timberwoof.spam-54...@nnrp-virt.nntp.sonic.net>):

Intel Towers as well.

George Graves

unread,
Jun 9, 2007, 5:58:59 PM6/9/07
to
On Sat, 9 Jun 2007 11:01:12 -0700, Timberwoof wrote
(in article
<timberwoof.spam-9A...@nnrp-virt.nntp.sonic.net>):

I use it, its way cool - especially if you have several apps (a mix of
windows and Mac) open at the same time.

gimme_this...@yahoo.com

unread,
Jun 9, 2007, 6:00:47 PM6/9/07
to
Just because you can't download the drivers doesn't mean they don't
exist.

MuahMan

unread,
Jun 9, 2007, 6:21:51 PM6/9/07
to

<gimme_this...@yahoo.com> wrote in message
news:1181426447....@o11g2000prd.googlegroups.com...

> Just because you can't download the drivers doesn't mean they don't
> exist.
>

LOL!

Sure you can use the Nvidia 8800GTX in your Mac...... in 4 years when the
11000GTX is out for Windows.

Timberwoof

unread,
Jun 9, 2007, 6:50:02 PM6/9/07
to
In article <466b11d4$0$8715$ed26...@ptn-nntp-reader02.plus.net>,
Jon Harrop <j...@ffconsultancy.com> wrote:

> Timberwoof wrote:
> > Parallels runs in a few modes, the most exciting of which is that
> > Windows applications run in the full Windows environment, but their
> > windows and UI appear entirely within the OS X desktop. The CPU to run
> > the Windows processes is not emulated; the software runs directly on the
> > Mac's own CPU. However, it uses main RAM to run, not video RAM. That's
> > generally reserved for things like video (and, in Windows, some printing
> > functions).
>
> Yes, fast-tracking a virtualized CPU is the easy part. Nowadays you can do
> it in hardware using Intel VT.
>
> Virtualizing the GPU is much harder and all other attempts that I am aware
> of have failed (except an alpha-release of Chromium for Xen).
>
> I have heard that Parallels does a great job on CPU virtualization, and I
> can well believe that because it is now a tried and tested technique.
> However, GPU virtualization is very new and very difficult. I believe
> Parallels are doing pioneering work here and I'd like to know how well
> their latest implementation works.
>
> You are correct that I did make several assumptions about their
> implementation. I assumed that it is not possible to do any virtualization
> on the GPU, so the task of virtualizing is moved either into the driver or
> into an abstraction layer behind the driver.

"The driver"? Ah. Windows does everything with drivers.

> Depending how this is done,
> the cost of that abstraction could be crippling, e.g. emulating PCI-E
> transfers.

OS X on PPC does an interesting thing: It runs the Classic Mac operating
system as a thread so that classic apps can run under OS X. And since
the 68k emulator is part of Classic Mac OS, this setup can also run
those old applications. One of my favorites was SuperPaint, which ran
faster in emulation under a virtual OS on my G4 than it ever did on
native 68k hardware.

> There is also the cost of sharing video memory between virtual
> hosts.

I still think you're making possibly wrong assumptions here. The 68k
emulator environment supplies QuickDraw calls to the applications, and
these are sent to OS X. There's no performance penalty for other
applications. (Except when running Word. Whether it's an old 68k
version, a PPC Classic version, or an OS X version, that seems to
swallow up CPU cycles whether it's doing anything or not.)

> Finally, high-performance video drivers are a hugely difficult
> software engineering challenge and require great sophistication. If
> Parallels somehow licensed nVidia's drivers and butchered them to get the
> best performance then they will almost certainly have introduced bugs.

Unless the Windows video system calls are tramslated into OpenGL calls.
Then none of that hacking about is needed.

Macintosh has a long history of providing everything an application
needs, as well as a long history of saying "this hack may work, but
don't do it for it may not be supported." It also has a long history of
applications that broke the rules and broke with new releases of the OS
... and applications that followed the rules and ran on several
generations of succeeding operating systems.

> I would not speculate as to how this is actually implemented or even what
> the performance trade-offs would be. However, I would like to see any hard
> numbers on how fast Parallels current implementation runs, particularly
> under high video memory usage (e.g. Doom 3 in Ultra quality on a 512Mb
> card).
>
> There are various articles out there but nothing I would believe:
>
> "Parallels doesn't officially support DirectX yet, but that didn't stop us
> from trying to run Quake 4 on it anyway. Alas, it was an exercise in
> futility--it did not work--and in fact, at one point, even forced a
> spontaneous system reboot!" -
> http://reviews.cnet.com/4531-10921_7-6546370.html
>
> This article doesn't seem to be very accurate but I can well believe that a
> virtualized GPU undermines the stability of all virtual hosts running on
> that physical host.

It does not seem reasonable to me to guess about the video performance
of native OS X applications based on the performance of a *game* under
emulation. If you want an indication of video performance, try this: I
upgraded my old G3/300 (ATI Rade 128 with 64MB of VRAM) with a DVD
player. Under OS X, that can play DVDs at 1024x768 without any dropped
frames.

If you have to run CPU- and video-intensive applications, then
fergoshsakes, don't run them in emulation under some other system. If
you want to develop OS X applications, then fergoshsakes, don't judge OS
X on the basis of its performance emulating Windows for games!

Timberwoof

unread,
Jun 9, 2007, 6:50:53 PM6/9/07
to
In article <1181426447....@o11g2000prd.googlegroups.com>,
"gimme_this...@yahoo.com" <gimme_this...@yahoo.com>
wrote:

> Just because you can't download the drivers doesn't mean they don't
> exist.

What is the difference between drivers that one cannot find or download
and drivers that don't exist?

Mitch

unread,
Jun 10, 2007, 1:57:05 AM6/10/07