Although I have been a lazy blogger lately we haven’t been lazy working on our remaining tasks at all. So, unsurprisingly, since my previous post we have reached and passed a few nice milestones. The latest one is that we’re finally able to build the gcc2/gcc4 hybrid Haiku images again, including all the software needed for the official release.
While that in itself isn’t a particularly impressive feat – after all we were already able to build the complete gcc 2 part before – the interesting aspect is how we are doing it.
Interesting progress for Haiku.
At last. Maybe my grandsons will see the release of Release 1.0 one day. Or grandsons of their grandsons.
I really think Haiku development will explode when package management is done / refined. Users and developers will finally be able to update the OS to the latest nightly build and install packages with just a few quick clicks.
Next big hurdle: Should the backend of the native API be replaced with Qt? (For those not following, yes, that is a serious consideration!)
That might be a good idea for the long term. Nobody is going to port applications to the Be API, as nice as it is, but there are many Qt apps and it is probably the dominant cross-platform API.
Sounds like a neat plan. Although I cannot exactly imagine how this will materialize. How will this differ from running QT apps in a QT runtime? Or will Haiku be the QT runtime itself?
Does a QT application have access to Haiku specific API’s like messaging/translators/etc? If so, you loose the benefit somewhat, since an application is not cross platform anymore and QT Haiku development still needs customization.
Edited 2013-08-23 08:38 UTC
Here is everything on the subject:
Where the discussion started (massive thread):
http://www.freelists.org/post/haiku-development/R1a4-initial-planni…
Follow-up:
http://www.freelists.org/post/haiku-development/QT-for-R2
Written idea (see the bottom of the page):
https://www.haiku-os.org/community/gsoc/2012/ideas
Follow-up:
http://www.freelists.org/post/haiku-development/GSOC-2013-QT-in-R2
Maybe we should just port the Be API over to linux, then get the linux people jazzed about it, call it … LinuxAPI … then they will start making new applications that are easier to port to Haiku.
Maybe a cross-development environment capable of supporting Linux, Haiku, and Syllable might be more appropriate?
Version 1 to support old Be apps of course. After that, hell yes to a Qt backend!
I think the focus should be on some usable applications such as an office productivity suite. I loaded the latest version of Haiku about a week ago and was pleasantly surprised that it supported all the hardware on my laptop. Then I went looking for some useful software, emphasis on the word “useful”. There’s a port of KOffice that is just a mess so I looked at Gobe Productive. The trial version from Haikuware looked promising but then I found out that Gobe has long since dropped sales or support of the package. My only recourse was to find a copy of Yellowtab Zeta, install it on a virtual system and then move the fully licensed copy of Gobe Productive from that virtual system to my real system running Haiku….what a PITA. But once I’d accomplished this I was quite pleased and had a working and useful system.
Haiku already has great hardware support and is quite stable. I haven’t experienced a single problem with it since I installed it over a week ago. But until it gets some useful apps it’s going to be a slow uphill battle for quite some time. Same can be said for AROS and OS4.
Updating the whole OS is way more important. Manually installing every new build is tedious and ridiculous.
Why bother to update it if there’s nothing to run on it? I know, it’s the chicken or the egg problem……but I think more people will actually load up Haiku and try it if there’s actually some apps for it and maybe even stick with it. Otherwise, what’s the point? Load it up just to look at the desktop?
I don’t think that’s being fair to what’s been done so far:
Play videos (any format), play music (any format), bit torrent downloads, download YouTube videos, a nice (but not HTML5 complete) web browser, email, instant messaging, image editing, audio editing, gcc / Bash / git / Vim / Python / autotools / IDE…
And those are all native applications, not including Qt ports.
EDIT: Flash drive support, WiFi, Java (!!), screensavers (I made one)
, web server…
Edited 2013-08-22 23:22 UTC
That’s true and I don’t want to downplay all the hard work and the quality of the native apps included in Haiku but it just doesn’t offer anything that cannot be found elsewhere. I’m not saying the Haiku needs a “killer app” to be successful either. I’m just saying that there’s no compelling need to even try it if I can’t do more than surf the web and play MP3s and videos. Heck 99% of the human race can do that on their cell phones. And I refuse to use Google Docs or other cloud-based services because I just don’t trust them or my govt. with my privacy. (I’m a US citizen btw).
I agree that porting QT, wxwidgets or even FLTK would be a huge leap for Haiku. I think we’d see a flood of apps on Haiku if this happened.
Edited 2013-08-23 01:38 UTC
Sure it does. It’s a free and open source operating system with the design philosophy of Mac OS X: Built together from the kernel to the GUI by the same unified development team, easy to use, 100% GUI driven NO command line necessary, easy to install and configure, unified look between applications.
On top of that it has a unique method of handling threads that prevents the user from EVER losing control (there is no concept of an hourglass icon or spinning beachball in Haiku).
Blah blah blah on and on I ramble. I’m sorry, my Haiku-fanboyism is really showing.
Nothing that will keep me busy more than an hour.
BeOS was very interesting when it appeared, nowadays it just doesn’t matter any longer except for people like us.
Of course not, applications just crash.
BeOS users remember how many applications suffered from thread coordination issues.
What? Those were my reasons why Haiku offers something that can’t be found elsewhere. What does it have to do with you being busy for more than an hour?
Edited 2013-08-23 12:10 UTC
Because I will spend one hour to:
– install Haiku into VirtualBox or similar
– play around with it while comparing it my old BeOS R5 copy
– decide I am better off with Windows and GNU/Linux
Nice project for nostalgic BeOS users, yes.
A good way for many OS hackers to learn about OS development in a multimedia oriented OS, yes.
I have all the respect for the Haiku developers, but there isn’t any single feature, now 13 years later, that makes it worthwhile to use as day-to-day OS.
I’ll leave cheap tricks for Windows family. Stay bitter.
There were a couple options you missed…
ThinkFree Office has been proven to run on Haiku using the OpenJDK port (with some hacks that make them launch from a launcher icon).
Google Docs would be a decent alternative as well – but I’m not sure the web browser is yet up to the task.
Why not just come out and say it: you really need/want OpenOffice/LibreOffice. They’re not easy to port, sadly.
Improving the web browser is more important I think – and will likely bring a lot more useful “application” support to Haiku. Don’t underestimate the usefulness of internet sites these days.
I agree about the importance of a web browser.
In regards to an office suite, I started thinking about this not too long ago. We have “StyledEdit”, which saves to RTF (open standard, Microsoft Word compatible). It might not be too hard to add image support, and then table support, and so on…
But then I came to a part where I got stuck: I think (this is just a guess) what people really are asking when they ask for an office suite is “Microsoft Office compatibility”. It doesn’t matter if it can do everything that Microsoft Office can if it can’t open the file formats.
So then I briefly looked into libraries that can read / write Office formats. There’s one very basic one (includes an API and command line tool) but is not nearly feature complete enough. It looks like the LibreOffice library is “part of” the suite, so if that was going to be used it would have to be used with LibreOffice.
Tricky. :/ Maybe the best course of action would be a native port of the LibreOffice GUI…
Good to read about the progress! I am still waiting and wanting to run Haiku on a daily basis.
But I wonder what the status of the kernel is. It is developed years ago and has not changed much the last few years. For instance I read news about the Linux kernel getting better power saving features, trim support etc. Is the Haiku kernel getting behind? Or are this typically features that should not be provided by a micro kernel?
The kernel has had thousands (millions?) of improvements since it was called NewOS.
Here’s the bug tracker on TRIM support
http://dev.haiku-os.org/ticket/7876
in a (pseudo) microkernel as haiku’s is, generally we don’t put this kind of support into the kernel. we usually prefer to run it in userspace, although it can be implemented at the kernel level (gotta love opensource).
GNU (never separate the two) Linux is a monolithic kernel, so drivers and kernel extensions run in kernel space, although it could be considered modular as well as some exceptions run in user space.
Monolithic versus micro, versus modular was bikesheded to death in the 2002-2005 Walter/OpenBeOS/Haiku era…