[openbeos] Re: openbeos Digest V8 #104

  • From: Christian Packmann <Christian.Packmann@xxxxxx>
  • To: openbeos@xxxxxxxxxxxxx
  • Date: Wed, 25 Jun 2008 13:20:26 +0200

Nicholas Blachford schrieb:
DarkWyrm wrote:
This could probably be a nice niche market for Haiku, considering its
low hardware requirements, if someone had sufficient interest in writing
for the hardware.

I think it's a wide open market so has good potential.

LOL. There seems to be far more OS competition on ARM than on x86, see http://www.arm.com/products/os/ .

Contrast that to the desktop where Haiku is not likely to ever be anything beyond a small 2nd league player.

Being a tiny 3rd league player on mobile devices would certainly be better? You are aware that Nokia plans to buy and opensource Symbian? How do you want to compete with that, in addition to embedded Linux and all the commercial OSs?

The problem is that most of these OSs are already tailored for mobile use in regards to UI and resource usage. Most will also have good IDEs and also commercial support (mobile Opera, maybe Flash plugins, etc.), which makes development easier. And even if companies have to pay for getting this support, this doesn't matter that much, as you already have fixed base costs for hardware development and production, so "being free" is not as attractive as it can be on the desktop. If you pay for your OS/IDE, but get better results with less man-hours, this will often be a better proposition than having a "free" OS which is more expensive during development.

Besides, even if you want to target the mobile market, x86 will be there soon. Atom and Nano are already well suited for so-called "nettops", in one or two process shrinks x86 should be suitable for PDAs/mobiles. Forking development to another ISA at this point in time seems superfluous; by the time Haiku is mature enough, x86 will be "good enough" for most mobile applications.

IMO an ARM port would be a waste of precious developer man-hours, which would be better spent on writing applications to make *using* Haiku attractive.

I don't mean that in a nasty way, but given that Linux has never even made a dent what hope has anyone else?

Linux is estimated to have 1.5% of the desktop market; the desktop market currently is about 1 billion machines. This gives us 15 million Linux users. If Haiku should eventually be 10% as successful as Linux is right now, it would have 1.5 million users, probably more than BeOS had at its peak of popularity.

You have to forget about the marketshare percentages and look at the absolute user numbers instead. The number of users determines how successful an OS is. More users -> more developers -> more apps -> more incentive for users to switch -> more users -> more developers -> etc. And I think that Haiku can be successful in the desktop market, even if it will never reach a marketshare of more than 0.5%. Once you reach a kind of "critical point", the OS will be healthy. Look at the Linux market; even though the marketshare is tiny, Linux development is still gathering steam. You don't need any kind of "final victory", just a sufficient base to ensure continued development of the OS and its applications.

Heterogeneous and Asymmetric processors rule the roost in the mobile world. However supporting these will give you an advantage going forward because that's the way PCs will eventually go as well.

They already did, years ago. Vias C7 and Nano have the Padlock engine and hardware RNGs. Most modern x86 chipsets support video decoding in hardware. Stream processing is getting widespread on x86 with CUDA/CTM; when GPU cores will be integrated on the CPU die, stream processing will turn into a standard feature. Intel will introduce AVX, AMD may introduce SSE5 if they live long enough. Etc. etc.

IMO it would be better to add support for such features to Haiku/x86 instead of dreaming about ports to other CPU architectures which are fighting a loosing battle against x86 in nearly all markets for user-interactive devices anyway.


Other related posts: