[haiku-development] Re: What's the status of Haiku?
- From: pulkomandy <pulkomandy@xxxxxxxxxxxxx>
- To: haiku-development@xxxxxxxxxxxxx
- Date: Thu, 21 Aug 2014 19:14:32 +0200
On Thu, Aug 21, 2014 at 06:39:55PM +0200, Julian Harnath wrote:
> Hello all,
>
> Ingo Weinhold <ingo_weinhold@xxxxxx> schrieb:
>
> > Unless I miss someone, of the Haiku developers (counting committers
> > only) who posted in this thread no one strictly opposed the idea of
> > switching to another kernel and most even seem to consider this an
> > interesting option.
>
> To be honest, I'm a bit disheartened by how many core contributors are
> expressing their complete willingness to replace our kernel with Linux
> (given that someone tackles the difficult work of pulling off such a
> switch).
I don't think I've seen "complete willingness" anywhere. Personally, I'm
interested in seeing how far this can go, and in the current state,
having this as one more way to make the BeAPI live, side by side with
Haiku.
> So, suppose for a moment we pulled it off and switched to the Linux
> kernel, maybe with Wayland for graphics on top. Great, so many drivers!
> What about integration though? One of the key benefits of BeOS/Haiku
> is, to me, the tight integration of system parts, from the lowest to
> the highest layers. Parts know about one another, talk with one
> another, react to changes in one another. Lack of that kind of
> integration is IMO the biggest problem that has held back Linux as a
> desktop OS the past 15 years.
> Could we provide the same kind of integration with a stack like
> Linux+Wayland underneath? And even if we can, we'd give up control over
> large amounts of the system to external entities. Maintaining a fork of
> Linux was not feasible back when Haiku started, and it still isn't
> feasible now. We would have to follow any kind of deep, big changes
> done to the Linux kernel, whether we want to or not. Same is true for
> Wayland, and possibly more components which all that would bring with
> it. We will not be the masters of what Haiku is anymore.
Let's see where this BeLinux project goes and how well it works. I won't
support a switch of Haiku to the Linux kernel if there aren't any proofs
that we can keep our system as nice and simple as it currently is. But
with someone (outside of our team of core devs) offering to do the work
towards this, it's not going to delay Haiku any further. So why not let
that project run, see where it goes, and review the idea of swapping kernels
when we can actually compare the state of the two solutions?
If it turns out an implementation of the BeAPI over Linux still can't
work as well as we want, we can very well wait 7 more years running our
current kernel, and maybe have another attempt then. As already
mentionned, there already were two similar projects in the past,
BlueEyedOS and Cosmoe. They both failed to gain traction. Maybe BeLinux
will do better, maybe not. If it does, we will have another BeAPI-based
platform to play with. This can only be a good thing for the "BeOS legacy"
ecosystem.
>
> In the past few months, whenever free time and motivation permitted, I
> have been working on this NFSv2/v3 file system driver for our kernel. I
> also had plans to do some more kernel work in the future. Now, I have
> to wonder.. is it really worth to continue on those things, if many
> Haiku devs already dream of a future without this kernel anyway? I
> know, nothing is being decided, I'm aware of that, but free time is
> limited and I don't want to spend it on something that will probably be
> completely dropped in the next few years anyway...
Or you can take this as extra motivation to show that our kernel can do
great things. And even support xattrs over NFS, which Linux still can't
do, for example. I don't think anyone should abandon development of our
kernel, and I will continue to support it. Let's prove to these Linux
guys that we are a serious project too and that we can do great things!
--
Adrien.
Other related posts:
