[haiku-development] Re: 64 bit

  • From: pulkomandy <pulkomandy@xxxxxxxxxxxxx>
  • To: haiku-development@xxxxxxxxxxxxx
  • Date: Wed, 10 Sep 2014 14:56:37 +0200

> > On Wed, Sep 10, 2014 at 04:44:53AM -0700, James Leone wrote:
> > > > Well, that's questionning the essence of the Haiku project. The main
> > > > goal of the project is "writing an open source binary compatible
> > > replacement for BeOS R5".
> > > 
> > > I know for an absolute fact that the corporate charter and stated
> > > charitable purpose, when the application for non profit status was
> > > submitted was:
> > > 
> > > To create a user friendly desktop operating system.
> > 
> > I can't find the words "friendly desktop operating system" in the
> > Haiku, Inc. certificate of incorporation:
> > http://www.haiku-inc.org/documents/haiku-inc_NY-DOS-Certificate-of-Incorporation-p1-of-3.jpg
> > http://www.haiku-inc.org/documents/haiku-inc_NY-DOS-Certificate-of-Incorporation-p2-of-3.jpg
> > http://www.haiku-inc.org/documents/haiku-inc_NY-DOS-Certificate-of-Incorporation-p3-of-3.jpg
> > 
> > Haiku, Inc. purpose is intentionally worded in generic terms to let the
> > Haiku Project (through developpers discussions and votes on this very
> > list) decide where the project should go. So I don't know where you got
> > that "friendly desktop operating system" from?
> > 
> > > 
> > > That clearly is not the primary goal of the project now. The primary goal
> > > cleatly has become to follow a plan that, as is clear from these
> > > conversations, a large portion of desktop users are actually adverse to.
> > 
> > We have seen maybe a dozen users raise concerns, some of them not even
> > strictly adverse to the plan, just questionning it. I think this is not
> > a large portion.
> 
> https://www.haiku-os.org/about gives the wrong impression then...
> 
> I don't understand why you're so defensive.

There has been several people joining this list lately, and starting
rather off-topic discussions such as this one here. This is the
haiku-development list, meant to discuss development issues and
day-to-day problems with Haiku. Questionning the project vision is
better done on other lists, the general haiku one, for example.

The result is a lot of noise on this usually quiet list and this makes
it difficult for developers to concentrate on what they are working on.

I think there is also a mixup of the goal (creating an awesome desktop
operating system, compatible with BeOS), and the means (the gcc2
compiler, the ABI stability constraints, the long release cycle).
Everyone wants to have their opinion on both, and most of the time it
seems for no particular good reason. It feels like people are often
thinking "32<64, so 64 bit must be better", or "2<4, so gcc4 must be
better". This seems to be a rather simplistic view of things. These
issues have been discussed many times, and repeating the same arguments
over and over gets boring.

So, here is what happens:
- Users question technical choices (32bit/gcc2) they shouldn't be
  worrying about (how does the choice of a compiler affect users? How
  does the choice of 32bit, which allows Haiku to run on more systems,
  can be a proble for them?)
- Our developers explain these choices and how they are related to our
  decision of binary compatibility with BeOS, and our choice of stable
  and long-term maintained ABIs, which are what makes the Haiku project
  different from other open source OS projects
- Then users start questionning these decisions, as it seems they are
  slowing down the project. No, they are not. Lots of efforts are put
  into those, because that's what the project is about. Should we
  replace these goals with others, lots of effort would still need to be
  made to reach the new goal. And part of the progress on the previous
  goal would be useless to the new one, resulting in even more waste of
  time.

There have been some valid concerns raised this time. 64-bit systems
might be useful in some very specific cases, which were mentionned
earlier: number crunching and high resolution video processing. If given
the choice between these (currently non-existing for Haiku) apps and the
support for old software and hardware (which does exist), I go with the
existing offer. Even if most of this software can be recompiled for gcc4
or 64-bit, as Stephan mentionned, it is better to do so when we have a
stable (and improved) gcc4 ABI to offer. The stable gcc2 ABI support
allows us to keep a stable system that can run all apps while we work on
the gcc4 side of things and make it better. Finally, we do have a 64bit
version of Haiku so people can start writing these apps needing more
than 4GB RAM each. It's now up to application developers to write such
apps, and make the 64bit version of Haiku more useful than the 32bit
one, at least enough to outweight the old software and hardware support.

I see no use in users, or 3rd-party developers coming from outside the
project, to just tell us "you are doing it wrong". People can disagree,
but for that to be useful they must take the time to explain why they
think we do things wrong, what a better solution would be, and what it
would bring us. I have seen many people telling us we should switch to
64bit, or stop using gcc2, but there are no actual arguments for doing
so. Finally, I must add that keeping 32bit and gcc2 support is not a
significant burden on our developers, the compiler works reasonably
well, gcc4 is available for things that really need it, and writing code
that works with both compiler os really not that hard. Sure, gcc2 lacks
C++11 and we would like to make use of that in some places, and we
started doing so in the 64bit and ARM versions of Haiku. But these are
purely technical issues, and I don't understand why users are so
concerned about them.

-- 
Adrien.

Other related posts: