Francis Charig responds to Opinion Article

[Waccoon]

I still wonder WHY Amiga Inc / TAO DID NOT release AmigaDE / AA for Amiga OS / Linux 68K.

From what I recall, Motorolla is trying to completely replace 68K with Coldfire and PPC.  Since Motorolla is an investor in TAO, that makes the possibility of a 68K port pretty slim.

I could be wrong.  Marketting attitudes change all the time.

By Kronos, from the original thread:

The DE is a good, but useless idea, cos noone
wants to let his SW run ony ANY HW without any
controll, and if you do controll the HW it runs
on like it has been done with the gamepack, you
could have just used a (maybe even free) portiblity-
layer like SDL,QT or similar and compile for
every target you want to support.

That's the choice of the developer, not a technical issue.  I don't see any reason why DE can't succeed.  Java seems to be doing well, even though it is architecture independent.

From what I can tell, TAO is a CPU translator.  Everything else is seperate.  All recompiling does is allow source code to run on a particular CPU.  Architectural differences and special features are handled by other things like the OS, drivers, the environment, and libraries.  Why is it such a far-fetched idea to have the CPU be translated, and let native drivers, libraries, and the environment handle everything else?

Also, the TAO technology DOES allow VP code to access special features of individual hardware.  It's up to the programmers to use it responsibly.

That's what I don't get about all this arcitecture indepenent stuff.  Once you allow programmers to use special features of a particular piece of hardware, THEY make the decision whether it will work on all platforms or not.  You might very well be using software that's based on architecture independent "DE", but then the application tells you, "Sorry, we only support computers with this particular feature".

You can't force developers to be completetly architecture independent.  It all depends on how many people and machines they want to support.  In my view, DE gives people the capability to support all machines, while still optimizing their code for individual systems.  It does not, however, guarentee that all apps written in DE will work across the board, allowing certain hardware platforms to succeed and others to be pushed out of the market due to a lack of support.  Idealy, what prevents this is the fact that software developers want as many customers as possible.  That doesn't work in the PC world, because there's so many PCs out there, and it's too hard and time consuming to do ports.  DE helps relieve the strain of getting software working everywhere.  I just can't guarentee independence.  Ultimately, that's up to the developers.

So, as usual, a complex marketting issue triumphs over technology.  Programs will be architecture independent only if programmers and managers want them to be.  That doesn't automatically make DE a bad idea, just a bit idealistic.

As someone who's been working with Perl for a while, now, I have no doubts that an architecture independent language can work and become very popular.  Hell, a lot of people say Perl is one of the best programming languages ever made (against my own opinion, of course).

I've always wanted an easy-to-use interpreted language like AMOS available on multiple platforms, because for the things I do, architecture independence is much more important to me than raw speed.  And supposedly, DE offers the speed of native code with the indepenence of an interpreted language.  If that doesn't fit the bill, you can always use Java in DE.

My only skeptisism about DE is whether Amiga can pull it off.  With so many financial rumors floating around, including those supported by Wayne, it's hard to tell where Amiga is headed.  The idea isn't the problem.  The problem is we need some kind of proof from Amiga is walkin on solid ground, that the existing community has a part in Amiga's future, and Amiga Inc. isn't just aiming for brand new customers.

So far, I am in no ways convinced.

[Waccoon]

The DE is a good, but useless idea, cos noone
wants to let his SW run ony ANY HW without any
controll...

By "his software", are you talking about the OS developer only, or everyone who writes software?  I don't believe application developers have a lot of hardware control.  A *LOT* of the software I use regularly at work is written in Java.  That's hardware control?

and if you do controll the HW it runs
on like it has been done with the gamepack, you
could have just used a (maybe even free) portiblity-
layer like SDL,QT or similar and compile for
every target you want to support.

I was under the impression you were talking about ALL architecture independent software platforms, not just DE.  My comments had to do with the idea of the DE, not Amiga's exact implementation.

Does anybody really know anything about DE besides the concept?  I'd really like to see this highly controvertial NDA and SDA, even though I have no intention of signing it.  I haven't had a lot of luck contacting Amiga after I bought the Party Pack.

The "run everywhere" won't help much with productivity-SW for several reasons:

Ah, so you are talking about the concept or architecture independence, not the specific implementation used by Intent and DE (which may change at any time until a final product is released).

a) There are only 5 platform to support.
Windows, Windows, Windows, Linux-x86 and OSX.

So, we might as well just give up on anything that doesn't have more than 5% market share, right?  I'm pretty sure there's a LOT more operating systems out there than Windows and Linux on x86, and not all of them a financial failures.

Intent in the DE-variant is quite cheap, but you have to sign a NDA&SDA before you can even start developing, and you are bound to pay a fee for
every single copy you produce by the SDA(*).

That might only be true of the SDK, not the full DE product.  I guess we won't know if Amiga Inc. changes the SDA / NDA until they are ready to release an official platform.  Bad PR and secrecy hardly helps their marketing situation!  But, so long as the "product" is in such a raw state, I guess they feel they shouldn't do it any other way... for now.

TAO did say that a lot of DE stuff won't work with Intent2, so I guess a lot of developing going on is experimental work.  I think it's a bit silly to release a product programmed for DE if the platform hasn't been fully standardized.  I'm sure a lot of stuff has to be rewritten anyway, every time a few changes are made to the platform.  The gamecards released a couple years ago were built on technology available at the time.

c) Apps done for a desktop, just DON'T scale to a PDA/mobile, so you will need to do a rewrite anyways (and there is QTopia to keep the QT-example).

Why not?  Mobile phones I can understand because of the sparce hardware, but PDAs are getting pretty powerful, with fast CPUs and reasonably sized screens.  Today's PDAs are a hell of a lot faster and more powerful than most classic Amigas, and we all know that a 14Mhz Amiga can still do some great stuff if it's programmed properly.  It seems to be more of an interface issue than a coding issue.  If you have to toss out all your work and start from scratch when porting from a desktop to a PDA, chances are you didn't plan your program well from the start.  Isn't that true of all multiplatform software?  If you don't do it right, just porting between Windows and a Mac can be a nightmare, and those machines are based on very comparable hardware that should scale just fine!

Besides, you're not going to write video editing software for PDAs.  How your software scales depends entirely on what you want to do with it.  Saying that software DOESN'T scale at all is going way overboard.

d) SW compiled with QT runs directly under the OS, and may even not be regognized as such, while DE/Intent looks much more like WinUAE or so.

So, what's wrong with that?  Does that mean DE apps can't talk to the OS at all?  If Intent can translate VP to native machine code, it should be able to translate platform calls to OS calls.  WinUAE hardly does any proper OS calls (at least that don't crash the host OS).  ;-)

e) Intent simple isn't competition to QT and
others when it comes to creating GUIs.

Well, I though that's what Amiga was doing!  DE is supposed to handle the GUI, multimedia, AVE, and stuff like that.  Intent is just the base platform, like Linux is just a kernel.  The interface is built on top of Intent, isn't it?

Thats all a bit different with games targeted to
low-end PDAs and mobiles, but honestly, I'm not
impressed with what has been delieverd sofar.

Neither am I, but then there's hardly anything to see BESIDES the game cards.  Like I said, I like the idea, I just don't know if Amiga can pull it off.  That shouldn't tell people that technology like DE is a bad idea, or that it is destined for failure.

But then, I still haven't seen the damn SDA and NDA!  I'm sure I'd think differently if I did, but only about Amiga, not the potential of DE.

What Would Amiga Do (Only Do It Properly)?