wrankin42 (+1.0)
9/9/07 3:39PM |
Wonderful!
I love RISC OS and want to use it all the time, but this is simply not possible. I've never had had that much cash and in any case my work demands a laptop. At present I have a venerable RiscPC dating from the last time I had real desposable income (aka student loan) and a relative's old Laptop running ROX.
ROX is good - it's gone from limping along to a pretty decent sim of the RISC OS way of doing things, over just the last few months - and it's a darn sight better than Windows. The only problem is that it's not RISC OS and can't run RISC OS apps.
Running VA on top of Windows is not satisfactory, even if I could justify a hundred quid for something that - no matter how many excuses I might make - I don't actually /need/.
ROLF looks like it has a long way to go but I think it is a welcome alternative to the current emulators.
Good luck Simon! If you get that Live CD going I'll be wanting one  ! |
simo (+1.0)
9/9/07 10:59PM |
so is this a WINE kind of thing (run RISC OS apps on linux) or some sort of "use a linux kernel to boot RISC OS" vmware-esx kind of thing?
the "look and feel" name suggests its more like ROX (risc os theme for xwindows) and then QEMU is thrown in (allows non-x86 emulation) to totally confuse me!
should be an interesting project, especially as varpc isn't going to happen on linux. |
Stoppers (+1.0)
10/9/07 5:37AM |
In reply to simo:
I've never used WINE, but ROLF certainly isn't an emulator, and there's no RISC OS involved.
There's also no X windows involved, the system uses the plain vanilla frame buffer, although the API allows for optimisation/acceleration transparently to the applications.
The aim isn't really to run RISC OS applications, although the behaviour of the GUI, as well as how it looks should be conducive to that (the GUI is co-operative, for example, but with true independent threads).
I've got a dozen applications running on this RiscPC. Toss in a couple that I'd like to have, mainly to do with digital video, and that's probably all the GUI applications I personally need to work with the system. To get them, I could write them from scratch, modify existing GTK applications to play nicer, or make it possible to run RISC OS applications, via re-compilation, emulation (QEMU) or BASIC (Brandy), combined with a RISC OS SWI compatibilty layer. It all comes down to which approach is least effort. |
jess (+1.0)
10/9/07 8:10AM |
Is the compatibility (or intention) such that the same basic (or any other interpreted language) app could run both on ROLF and RISC OS? |
Stoppers (+1.0)
10/9/07 3:11PM |
In reply to jess:
Yes, that is the idea. Brandy can fairly easily be modified so that the SYS command makes a call to a library routine that can be provided by ROLF. I expect QEMU can be just as easily modified, for compiled applications.
The task then boils down to mimicing a useful subset of RISC OS system calls (SWIs) so that real applications can be run.
While there are many system calls that would have to be implemented to make up a full environment, many, especially the lower level ones, won't be used by the kind of user application that people are likely to want to run so, hopefully, it's not an impossible task.
|
lym (+1.0)
10/9/07 4:25PM |
Is the Riscose project of any relevance to this? From memory, they were trying something at least superficially similar: emulating SWI calls to enable RISC OS applications to work on top of Linux. If their sources are still available, might they be helpful? |
jess (+1.0)
10/9/07 6:05PM |
Would it be able to load its own versions of modules that provide more swis to the system? |
Stoppers (+1.0)
10/9/07 9:44PM |
In reply to lym:
Riscose doesn't seem to have been touched in about five years. However, the most recent update does sound very much like it could be useful. (As could Matthew Bloch's web hosting service, bytemark).
The difference is that the major part of their work (as far as I can tell from a quick look at their site) was on the ARM emulation and, apparently, BASIC, too. These have been superseded by QEMU and Brandy, and the SWIs I'm most interested in are more WIMP oriented than command line.
There's also the open source shared C library by Graham Shaw, which will probably will come in useful.
In reply to jess:
Do you mean RISC OS (ARM code) Relocatable Modules, or a ROLF native system? For the former, it's probably possible using shared memory and some synchronisation between instances of QEMU; for the latter, I think shared libraries are probably the way to go in most cases.
Bear in mind that, sitting on top of the Linux kernel, applications can use pre-emptive threading and non-blocking I/O, so many of the uses for RMs are not needed under ROLF. |
jess (+1.0)
10/9/07 10:08PM |
I was more thinking in terms of something for ROLF itself. So if a different programmer wanted to make something work that needed an unsupported SWI they could add it in a manner similar to RISC OS modules. Being able to use real RISC OS modules would be good for compatibility, but I assume much harder to make work. |
lproven (+1.0)
11/9/07 1:13AM |
It's a very interesting idea. It would make a unique sort of Linux distribution!
Actually, a project like this has been attempted before: to build a BeOS compatible system on top of Linux. It was called BlueEyed OS, originally just BlueOS. I believe it didn't get very far. :¬( Haiku is the whole thing from the ground up - clean reimplementation, like ReactOS is for Windows - and it's doing much better.
Less of a new OS, more of a sort of turbo-emulator, was Amithlon. It's dead now for legal reasons but essentially it was a small custom Linux distro that booted a PC straight into a Just-In-Time compiling Amiga emulator and thence straight into an Amiga desktop. Essentially it was a boot CD that turned a PC into a fast Amiga. Very nice if you like that sort of thing!
(This is based on a post from the ROX mailing list; I just thought it might be interesting here.) |
Stoppers (+1.0)
11/9/07 6:12AM |
In reply to jess:
Well, that should be the beauty of open source; the programmer who needs a particular SWI can simply add to the compatibility library. If they want to provide new native ROLF/Linux features, they can provide a library for applications which want to use it (and it may get picked up by other GNU users, if it's generally useful).
Using RISC OS modules would be possible, I suppose, it will just come down to whether anyone feels a strong enough desire for it (as with pretty much any feature). |
Stoppers (+1.0)
11/9/07 6:27AM |
In reply to lproven:
I'm hoping that this project won't meet the same fate!
My USP is that I'm not setting out to provide a comprehensive RISC OS environment, that is just a side effect of producing a RO-like interface, so I can't be accused of failing just because somebody's favorite program doesn't work on it.
On the other hand, success would be having established RISC OS application developers selling versions of their software specifically for ROLF. |
mripley (-0.1)
11/9/07 8:59AM |
Wouldn't the time be better spent getting Linux to adopt certain RiscOS methods such as the way applications are organised or the way "hidden" files are not needed.
I have just recently installed PCLinuxOS on a laptop and have been very pleaseantly suprised (I gave up on Ubuntu due to issues with my hardware, nVIDIA drivers!, and the need to start using the command line). I would recommend PCLOS to anyone looking around for a RiscOS alternate, it even has a window decoration called RiscOS as part of the distribution! Granted PCLOS is meant to look and feel like windows due to the need to
attract those users. However you can easily change all that and right now my PCLOS desktop looks almost the same as my RiscOS one.
The biggest issue I have found with PCLOS/Linux is all those damned hidden files in the users home directory. It's exceptionally messy and very dangerous if non enthusiast users start to adopt Linux. I think this is an area that RiscOS's approach to organising files, folders, applications and configurations can be of great use.
As for applications themselves I don't think there is much to be gained. Some of the smaller apps maybe. A simpler cleaner FTP certainly! The default one is clunky and looks all the worse for having all those hidden files visible! PCLOS/Linux already has a wealth of apps with features that are sadly lacking under RiscOS (a reason for me switching). For example video works! here's a list of some of the apps that come as part of a PCLOS distribution:
Gimp (Image editing)
Thunderbird ( email and news) - Excellant "new/unknown" message handing
Firefox (browser) - version 2 as opposed to RiscOS v1
OpenOffice - office suite that can also read/write M$ if you are peverse enough 
Calc - spreadsheet
Writer - wordprocessor
Base - database
Draw - ! yes
Impress - slides multimedia presentation
Amarok - music player
Beryl - 3D desktop manager
Frostwire - peer to peer
MPlayer - video/audio player
FYI everything worked "out of the box". Network setup was perfect and automatic. Wireless connection to my Canon printer served from XP laptop, worked first time. All in all an astonishingly pleasant non windows and totally free experience.
Remember folks all of the above is FREE.
So my suggestion would be to build on top of what is already working and make it better. A RiscOS (PCLOS+ROX+?) style distribution. |
Stoppers (+2.1)
11/9/07 12:09PM |
In reply to mripley:
"Wouldn't the time be better spent getting Linux to adopt certain RiscOS methods..."
How would you suggest I go about that?
I don't think just saying that people should try a different approach will have any effect, they have to be shown it and then maybe they'll decide that it is worthwhile. The KDE developers weren't even convinced that right click on scroll bar buttons should work in the reverse direction to left click!
What I'm doing is showing another way of using Linux to whoever is interested. If people like it, maybe they will adopt it, but it's more likely I'll just end up with a unique desktop system.
The hidden files in the home directory are the standard Unix approach to storing choices (and they are hidden, after all). You can make the same argument about non-enthusiast users shift-opening application directories.
|
mripley
11/9/07 6:37PM |
Simon,
Demonstrate it with a distribution. It does not have to have all the bells and whistles applications since the aim is to demonstrate a tidier more functional approach.
Shift-opening applications is a very deliberate action whereas the hidden files in /home/<user> are not really hidden are they? The are visible depending on which application you use, for example KFTPGrabber and Krusader. With these apps you are free to browse....and affect these hidden files....and thus have an accident even though the original intent had nothing to do with folder in question. This is why I am 100% convinced this is an accident waiting to happen (i.e. a very non computer literate user weened of Windows). Just because something is standard does not mean it can't be improved upon. |
jess (+1.0)
11/9/07 6:44PM |
Several years ago I subscribed to a mailing list that was going to create a ROX-linux. What happened was major discussions on how to change ROX filer, I don't know if anything ever got released. I think Simon's approach is far better. I think the only dissapointment would be if there's no compatibility with ROX applications (or at least a simple port being possible.)
I think this will be of interest to all RISC OS users who have a PC and those who want a laptop. |
Jaffa (+1.0)
12/9/07 9:57AM |
See also http://www.bleb.org/software/wimpos/ - a ROX AppDir based Linux-kernel-running OS which I was starting it. |
Stoppers (+1.0)
12/9/07 10:09AM |
In reply to mripley:
"Demonstrate it with a distribution."
So you're saying I should prepare a live CD, or something like that? |
DS1
12/9/07 5:21PM |
And base it on GOBOlinux which already has a very riscos like directory structure anyway. |
Stoppers (+1.0)
12/9/07 6:58PM |
In reply to DS1:
Thanks for your suggestion, but the live CD is based on Linux From Scratch http://www.linuxfromscratch.org/ - which means I have complete control over where all the files go. |
mripley
13/9/07 8:26AM |
I'm a great believer in not re-inventing the wheel....unless it's square! So take what already exists and works well and improve it. I get the distinct impression from the Linux world that Ubuntu is being hyped up and PCLOS is gaining ground due to its flexibility and greater hardware compatibility. A darn sight easier to configure as well I might add (I abandoned Ubuntu). So my suggestion is for a PCLOS distribution with RiscOS improvements, especially the location of those pesky hidden files. RiscOS sticks them all in !Boot, which is far far superior home for them (or should that be /home )
Mind you I noticed somebody mentioned GOBOlinux, I need to investigate.
Also note you don't want some distribution that lives amongst the enthusiasts. You want to affect the mainstream. We know RiscOS has features that are far superior to the existing 3 so wouldn't it be good to have those features adopted by linux-world ?
|
mripley (-1.0)
13/9/07 8:49AM |
Ok I took a look at GOBOLinux and it looks like the solution is already there. There are a number of flavours that people have built which brought together into one distribution would be what is needed, specifically:
1. A filesytem organised more tlike RiscOS (they actually based it on Macs)
2. Somebody has GOBOLinux with ROX (riscos filer)
3. Somebody has Gobolinux on an Xscale (!!!!)
4. Somebody has GOBOlinux with beryl
Number 4 presumably means that compiz-fusion should be possible. All of this would enable you to have a very RiscOS functioning desktop and filer. In addition compiz-fusion would knock the socks of any presentation that Windows and Macs can muster.
So it seems to me that the starting point could be GOBOlinux with ROX and then get RiscOS apps added to that distribution. |
Stoppers (+1.0)
13/9/07 1:37PM |
In reply to mripley:
Thank you for your advice, but, I think I'll carry on as I was.
I'm not setting out to "affect the mainstream", I'm working to produce a system that's as close as I can get to RISC OS, but with cheap and fast hardware, a more flexible kernel providing access to programming features, libraries, peripherals and, to an extent, programs that aren't available or even possible on RISC OS.
I'd also be happy if it provided a way forward for RO users and companies, but that is certainly not my goal; I'm just scratching an itch and making it available to anyone who wants it.
By the way, the concept originated with this conversation on the Iconbar, back in May 2003:
http://www.iconbar.com/forums/viewthreadedlist.php?threadid=3945
If it was just a matter of configuring a Linux system, that's what I would have done. |
wrankin42
15/9/07 2:19PM |
Any chance of a ROLF news RSS feed?
In anycase I hope Drobe will keep us informed of any future ROLF developments.
|
Stoppers
16/9/07 10:01AM |
<censored>in42:
No, sorry, web site design is not my forte (look at my website!).
I wouldn't know how to set one up, and I'm concentrating on getting a single script to generate the live CD at the moment.
Unfortunately, for personal reasons, things will be going slowly for the next couple of weeks.  |
JanRinze
16/9/07 2:08PM |
Hi Simon,
I have tried the code from your website.. it shows a background image and a mousepointer (which moves when I move the mouse ) but that is all I can do with it..
Jan Rinze. |
Stoppers
16/9/07 5:00PM |
Argh!
Sorry, those who e-mailed me; I messed up my e-mail configuration and didn't get any mails!
It's working, now. |
Stoppers
20/9/07 2:31PM |
Thanks to Ned Abell, who has kindly provided storage space for the ISO, the first version of the live CD is available via my Drobe website, http://stoppers.drobe.co.uk/
It should give you the basic GUI and some applications which work to a greater or lesser extent.
I'd appreciate any comments or suggestions (but I already know it has to be more stable). |
| Please log in to post a new comment |
An ambitious initiative to eventually run RISC OS desktop software on Linux-powered PCs took a step closer to reality this month. Issues with the user interface code have been addressed, and a bootable CD is being prepared to demonstrate the system's abilities.
