South West 2005 show report - RISC OS News, Software and Information

South West 2005 show report

Published: 28th Feb 2005, 03:12:07GMT Source: drobe.co.uk
By Martin Hansen

Page 1 of 1

Coverage of Castle, Qercus, UPP and ROL theatres

Martin Hansen returns from the West Country, having learnt of future RISC OS projects in the talks given during the South West 2005 show. How have times changed since last year's show?

This year has been a fairly quiet one, so far, for RISC OS. This is not, necessarily, a bad thing. My feeling is that there are currently several exciting projects in prospect but that developers need more time and a continuing calm background environment against which to perfect their next offerings. So, it was with no great expectations of any big breaking news story that I zipped along to The Webbington Hotel in rural Somerset to, once again, report for drobe.co.uk on the opening show of the RISC OS year; the South West show 2005.

I took along my Mathmagical Software Company stall but, without apology, largely left it to run itself in order to attend the four talks of the day, from Castle Technology, Qercus magazine, the Unix Porting Project, and RISCOS Developments Ltd. Before detailing these talks, however, I must comment upon how good it was to see that the show had attracted a full compliment of exhibitors, and pulled in a comfortable number of visitors. Although I do not feel able to describe the show at any point as being busy, from what I saw, and asking around, the show never felt dead or empty; a calm show but nicely so, was my impression of the event.

Castle and Tematic logo

Castle
Castle Technology's managing director, Jack Lillingston, kicked off the theatre presentations with a gentle and good natured demonstration of the products his company are currently pushing. They include, of course, the Iyonix, along with a USB 2 card reader, a high quality Canon digital camera, a Canon photographic quality printer and an Epson scanner. The talk was similar to his presentation from the previous year. Jack showed us the camera outputting to the printer directly (once he'd remembered to turn the printer on) and also how one can simply take the memory card out of the camera and feed the images into an Iyonix via the card reader. The quality of output onto photographic paper was passed around - it was stunningly good. Using David Pilling's scanning software we saw how fast and smoothly the scanner works through USB 2; about two and a half times faster than with USB 1. Again the quality of the image was very crisp.

It was interesting to see Jack using not Thump as in the past, but instead Rob Davison's Variations. I've heard several people enthusing about this recently, usually along with a comment that they cannot believe Variations is free to download over the Internet.

Judging by the way the audience hung on to Jack's final brief about Castle's immediate future, we're all very curious to know where Castle are taking RISC OS 5. Due to apparent non-disclosure agreements, the clues are all fairly cryptic, but let's work our way through them: Tematic, a trading division of Castle Technology, have been hiring software engineers since just before Christmas, and are likely to be doing so at certain 'key points' throughout 2005. They are said to be moving to premises with five times the floor area of the existing set up. The number of big projects being placed with Tematic for development, and the variety of companies interacting with Tematic, is growing steadily, the audience was told. All of these are, inevitably, of an embedded nature and, so far, most of the resulting products are not sold in this country. In fact, the end user is unlikely to be aware that RISC OS is involved at all although Jack did say that the main giveaway clue to look for is the high quality anti-aliased fonts in a product for the home. You now have 30 seconds to crack the 'Castle conundrum'.

It was interesting to hear Jack run through the RISC OS unique selling points; a software stack that is arguably stable, reliable and robust, and one that has evolved and been debugged over a long period of time. It's here that competitor companies, formed more recently, struggle. Castle are very clear about not wanting to compete with Microsoft in the desktop market. We'll get spin-offs from the embedded work - USB 2 being a perfect example - but Microsoft has too much money and resources to make taking him on in a direct way anything but folly. This may not be 100% what we want to hear in our fantasies but it does make good, realistic sense. A warm round of applause concluded Jack's talk; with everyone wanting Castle to succeed even if the desktop market is not, and can not be, the main focus for the time being.

Qercus logo

Qercus magazine
The next speaker was John Cartmell, editor in chief of Qercus magazine. I think he'd have been given a hard time had the sixth issue of Qercus not popped through letter boxes a few days before the show, albeit a few months late. The main problem John and perhaps the rest of us are having to grapple with is a forked OS, and two companies that, foolishly, managed to fall out big-time and in public half way through 2004. Qercus, inevitably, is for desktop RISC OS users but we are in this curious situation where the manufacturer of our best machine is having to focus upon embedded work in order to expand and, furthermore, does not, and can not, currently ship that machine with the most feature rich, enhanced version of the OS. Without doubt, Iyonix owners are a band of dedicated RISC OS followers, but (it's believed) RiscPC users still vastly outnumber them and, one presumes, purchasers of VirtualRiscPC currently form the biggest area of growth in the number of new and returning users to our platform. The enthusiastic amongst these virtual and real RiscPC users must be working with the Select and Adjust versions of RISC OS on a daily basis.

From his talk I concluded that John must have mulled over these issues as much as anybody. While not wishing to snub Iyonix owners in any way, he has concluded that it is time for Qercus to run some articles on the features contained within the latest versions of RISC OS 4, and so we can look forward to tutorials on what the new Paint and Draw can do. He is clearly optimistic that the A9home, whatever that turns out to be, is going to be a major hardware advance within the market as will be the 32 bit version of RISC OS that will have to be shipped with it - Qercus will be following and supporting developments here.

John then detailed mostly what one would expect; a mixture of articles for the beginner and then for the more advanced user, plus a return of focus to the programming side of computing, and a continuing flirtation with computer art. Personally, what I do like about Qercus is that it is a magazine you want to keep in order to refer back to it's features and article at a later date. Acorn User of old had very much become a read and then throw away magazine. So, we are supporting you John: A little more quantity at the same quality, please.

UPP logo

Unix Porting Project
The third talk of the day was from Peter Naulls. This drew a large audience, a little different from those who attended the other talks of the day due to a good number of the younger enthusiasts dotted around. The ideas being batted about by his Unix Porting Project are exciting although it has taken a little while for them to be properly grasped by the average RISC OS user, myself included. Briefly, the plan is to set up a pipeline through which a hoard of Unix or Linux applications can be fed such that they emerge as software that will run, stand alone and natively, under RISC OS. The idea is ambitious but the reality is that Peter can now demonstrate various pieces of software that have passed successfully through this process.

Some fascinating issues emerge; it is certainly amazing to see these packages working on an Iyonix but they don't look like RISC OS applications. Some come with pull down menus or non-standard window furniture. Some games work, but assume that more processor power is available than the XScale can deliver. One member of the audience voiced the opinion that he thought it was quite reasonable that a Unix application should continue to work, respond and feel like a Unix application when running on RISC OS. Peter explained that, in some instances, part of the conversion process could replace, for example, the Unix menu system with the RISC OS pop-up one.

The real excitement currently comes, however, from the possibility that a state of the art web-browser could be ported into RISC OS. Peter listed twelve RISC OS browsers, all of which are brave but unsatisfactory efforts to give us this competent and killer piece of essential software that we still (embarrassingly) badly need. FireFox is the object of desire, and porting this across has been the principle focus of the Unix Porting Project for the last couple of months. With a wry smile Peter offered to show us how well the latest port runs and then crashes: If you blinked, you missed the bit where it ran. However, it truly will be a great day if this can be successfully moved across to our platform and tweaked to work at a good pace under RISC OS. Peter got a very positive round of applause at the end of his talk; I hope it spurs him on.

UPP logo

RISCOS Developments Ltd
For the final talk of the day, Paul Middleton, of RISCOS Developments Ltd aka RISCOS Ltd, took to the floor. The rate at which Select advancements have been released has slowed of late but this is not, Paul assured us, due to any diminishing of effort on the part of RISC OS 4 programmers. Rather, Paul said they have been grappling with the efforts of, one-by-one, making each of the 171 component RISC OS modules 32 bit neutral. With good humour, Paul talked of the Acorn rushed patches and bodges and quick-fixes that had been uncovered as his company finally grappled with certain modules that they had left untouched for as long as possible, mainly because no-one was left around who fully remembered or understood the original code.

The goal is to complete Select 4, due out this summer according to Paul, which will be the first 26 bit version of RISC OS to be derived from the newly constructed 32 bit neutral source. It's easy to become confused by this issue and so Paul carefully explained that although RiscPCs could, in principal, run the 32 bit neutral source there would be no point as a lot of 26 bit software would then need to run under 26 bit emulation. The same 32 bit primary source would ultimately provide the 26 bit version for RiscPCs, a 32 bit version for hardware such as the A9home, and a 32 bit version for hardware such as the Iyonix. Alas, the Iyonix is at the end of the chain and the part of the plan that is likely to be the hardest to justify from a financial point of view.

As a user with the Adjust ROMs I paid close attention to the standard part of Paul's talk concerning how to boot from alternative sources such as a CD ROM or a network. The key step is to fire up the computer with the Shift key held down. I've just tried it, and sure enough, up comes a green BootMenu screen offering such options. Hard disc problems and issues formed a good sized chunk of Paul's talk, and he has clearly had some bad experiences with hard disc failures taking out valuable data as they die. Like most people in the room I began to feel a little uncomfortable. Yes, I would lose quite a lot of photographs and documents if my Kinetic hard drive failed right now; one of the reasons I've stopped bouncing it around in the car by taking it to shows quite aside from not wanting an "old" machine on my stand.

A related concern is that the latest hard discs capacities exceed the capabilities of accessible permanent back up storage. Unless networked to a file server, Paul argued that a user could quickly fill up a hard disc with information and media beyond what can be conveniently backed up. The proposed solution is to introduce partitions that prevent the user storing in one place more than can comfortably be backed up by, currently, a CD-ROM disc. Another issue concerns being able to rescue a hard drive from a failed RiscPC by plugging it into a Windows PC. Presently, you are unlikely to be able to easily pull any off any individual files under Windows. The solution here could be to format hard disc in a standard Windows recognised format, with RISC OS being designed to work with this format rather than with it's own specialist hard disc format, as in the past.

For me, Paul's talk really took off when he began talking about how mind-bogglingly powerful PC graphics chips now are. Typically, they are capable of drawing millions of polygons per second. In part, I presumed these observations to be tied in with Project Simon although he did not refer to it as that. Reading between the lines, this seems to involve some sort of graphics card, possibly on a podule like a Viewfinder, with one of these chips on board, maybe for a RiscPC, in which the modified OS uses the ARM CPU to manage the graphics chip rather than the current approach in which our slow ARM processor generates the graphics itself and shunts the resulting images around slow internal buses. This enhancement at the operating system level with a hardware add-on seems to go, hand in glove, with the possibility of replacing all bit-mapped parts of the desktop with vector graphics equivalents. Can our ten year old RiscPCs really be on the edge of having an option that revolutionises the nature and speed of their graphics capabilities?

With the four talks still buzzing around my head, I packed up my stall into the car and headed for Pizza Hut to talk some of the ideas of the day over with my eighteen year old son, James. It was his first RISC OS show and he had enjoyed the talks and walking around. It is good that the RISC OS world is still alive with ideas, and people who are responding to the challenges facing the quirky British operating system that refuses to die. I conclude by thanking organiser John Stonier for, once again providing a forum that yielded an interesting and thought provoking day. Already, I am looking forward to the Wakefield show. So much is now simmering that surely that is when we'll be able to actually buy the products containing some of these ideas.

Links
Back to the front page
Peter Naulls and Martin Hansen are both drobe.co.uk writers and show exhibitors

This article has been linked to, or is available in the following formats:

[Printable] [Digg this] [Blog search]

md0u80c9 28/2/05 6:56AM	Great article Martin; thanks for all of this. Particularly welcome for those poor saps (like me) on nights, so physically unable to make the show.
DaveW 28/2/05 7:04AM	Thanks Martin - a very absorbing article. Very, very disappointed to hear that Select 32 for Iyonix is so far down the line - is that political, or are there really so few iyonixes? It is a superb machine - it just needs Select!!
rmac 28/2/05 7:34AM	Good summary for those of us too far away to attend. Thanks Martin. Pity about Select 32 for the Iyonix though.
dgs 28/2/05 9:07AM	*In reply to DaveW:* "are there really so few Iyonixes?" Some people would like you to believe that. Most of the available evidence, though, suggests that it's far from the case. To put it in simple terms, there are plenty of RISC OS users who own Iyonixes, but also plenty of RISC OS users who don't. dgs
Eddie 28/2/05 10:28AM	A good read - thanks Martin. But adding a graphics card to a 10 year old RiscPC does seem a strange business model to me... but what do I know?
benc 28/2/05 11:57AM	*In reply to eddie:* If it is related to project simon (which seems likely) - then my guess would be it is related to the A9 not the RiscPC.
JGZimmerle 28/2/05 12:13PM	@Eddie: Surely modifying the OS to use the GPU (graphics processing unit) for most of its graphics stuff is a good idea. If this would be usable on old systems like the RiscPC, then this would be a nice side-effect. MacOS did this some time ago and IIRC it resulted in a faster GUI and a faster system. CPU and main memory resources previously used for graphics were freed up while the desktop made much better use of the modern GPUs, wich previously were largely idle in desktop use.
johnpettigrew 28/2/05 12:14PM	As for the issue of changing disc format, please tell me they're not going to use FAT32? It's a horrible format. It would be better to use NTFS but Microsoft don't tell people how that works without lots of cash - GNU/Linux has only gained proper read/write NTFS access in the past year or so. Of course, one good solution would be to use one of the linux formats, such as ext3, which would give us an efficient, journalled filing system with proper long file names that would still be accessible from Windows boxes (because there is software to read ext3 on that OS, too).
JGZimmerle 28/2/05 12:16PM	Of course to make complete backups, you could also simply put your ADFS formatted HDD into a Linux PC and make a compressed disc image.
Eddie 28/2/05 12:50PM	Hi Julian. I agree, the idea is a good one in principal, but I just wondered what the business case was for a RiscPC podule. If it's for the A9 then it makes more sense but I wonder how long we will have to wait for that? In the meantime I have a shiny new Mac to play with...
sa110 28/2/05 1:33PM	All I can say is "thank you" to the Unix Porting Project for doing their best to port such an upto date browser such as FireFox, and may you work at god's speed(if you beleive in such an entity, failing that, let's just settle with the sooner the better) to bring it to us. Oh and, everyone rush out and join the UPP!
sa110 28/2/05 1:35PM	The article mentions ROD Ltd, but their website is still ROL.
mavhc 28/2/05 4:18PM	Either PM isn't good at explainging things, or people aren't good at reporting things he says, or he's gone mad.
simo 28/2/05 4:37PM	Maybe I'm missing something here, but isn't the obvious choice of backup just to put two drives in the machine - one working, the other for backup (you could even unplug it if you're a power consumption freak)? I've been doing that for years! Basically as CD, DVD and tape are not big enough to backup the 400Gb drives we have these days, the only sensible option is to just have another 400Gb drive. Either that or use a Linux/Windows fileserver and just backup over Samba/FTP/NFS (again, we don't need to know about ADFS when we're on a network share). I think the NTFS idea is ridiculous - Linux has only experimental write access. And I don't think Linux can read a whole ADFS formatted drive, just floppies - and I doubt it's F+ format, think it used to be D?
mrchocky 28/2/05 4:52PM	That'll be why Linux has no problem accessing 80GB ADFS formatted drives, then? Perhaps a little bit less guessing is in order.
fwibbler 28/2/05 4:57PM	While I'm all in favour of a more up to date filing system, I hope that the way that Windows fragments files at the drop of a hat can still be avoided. Can it? Cheers!
robert79 28/2/05 5:03PM	If you had a failed hard drive in RISC OS format, surely you could just plug it inot a RISC OS machine to try and rescue rather than a Windows machine? In fact I don't see why the format matters at all - if the drive's fscked it's fscked...
diomus 28/2/05 5:28PM	*In reply to robert79:* What you say is true, however the article talks about what to do with a RISC OS formatted hard disc if your /RiscPC/ fails.
robert79 28/2/05 8:53PM	Oh, teaches me for skim-reading! Still of limited use to read it on a PC anyway, unless you had an urgent need ot get at your collection of MP3's or something.
mavhc 28/2/05 9:37PM	You plug it into your spare RiscPC you got off ebay/from your local school. Whilst ADFS/Filecore isn't brilliant it's good enough (Although >2GB files would be nice), backup your files to your PC's HD if you care, and from PC to RISC OS too. The only real concern about RISC OS harddrives is the max size, IDE and Hform limited to 128GBs, ADFS/Filecore limited to 256 with a 512byte sector size. Castle have updated HForm for 256GB max, but the second 128GBs isn't DMA'd due to VIA's ATA controller, things were much better when we designed the whole system. Of course you'd only need that much space once a decent video player is released. Get DVD writing working and that's 4.3GBs, or 7.9 on dual layer, enough for most people. This thing about partitions sounds someone's got the wrong end of the stick. Iyonix serial number IYO1105019 so I estimate there's 5019 iyonixes
govind 28/2/05 11:21PM	*In reply to mavhc:* Hmm... how long have you had your Iyonix?
dgs 1/3/05 12:44AM	This discussion about capacities of different media is silly. I've been using 400GB (compressed) tapes for over two years now. At work we have nearly a thousand of them. The next iteration is 800GB, and double that is not far off. dgs
Gulli 1/3/05 1:05AM	I have to agree with johnpettigrew about the disc formats - for goodness sake don't use FAT32! ext3 would be a far more logical choice since all the info regarding that is freely available and it is quite a reliable format.
simo 1/3/05 1:41AM	every os in the world practically can read/write fat32, how many can even read ext3? and if you want an up-to-date and open filesystem, why stick with ext3 and not reiserfs, reiser4, xfs, jfs....
mavhc 1/3/05 3:34AM	400GB compressed, any home user with 400GB of stuff will mostly have video that's already compressed, so that's maybe 250GB. A 250GB HD costs 100ukps, a 400GB tape drive costs 1500ukps. Tapes are 35ukps. Number of people buying tape drives: 0 Hopefully Bluray will provide a decent solution, meanwhile I'll stick to having copies of stuff all over the network and writing DVDRs. As we can see there's no suitable FS for being able to read a HD on Windows If you want to read your RISC OS harddrive on it then use an emulator. *In reply to Govind:* Dec 2004. Did Archive Magazine ever publish their Iyonix serial number experiment from Apr 2004?
Eddie 1/3/05 7:47AM	*In reply to mavhc:* I dont think the serial number you quoted is very meaningful. My Iyo is 18 months old and is IO1200983. The internal number quoted by HID is probably more relevant, or even the network card number, which has to be unique.
mrtd 1/3/05 8:57AM	Dis I read the report on PM's talk correctly, and did he really suggest limiting disc partition size to that od a CD ROM (700MB). I don't like this at all, it would be horribly restrictive. You would be for ever running out os space in partitions and having to rearrange things. No, I can't believe he really suggested that. I too don't like the idea of adopting a PC file format. You can, as mavhc pointed out, already read RISC OS formatted drived inder Windows, should you be perverse enough to want to do such a thing. I much prefer to have a filing system that doesn't fragment files. Martin
hzn 1/3/05 9:33AM	As for harddisc: Since Linux can reas ADFS you can put the RISC OS disc into a Linux box (or perhaps a PC booting Knoppix debian Linux off CDROM) - but the question arising is if the disc is understood since normal PC do expect a specific kind of how partitions are defined on the disc - even if it is just one partition as in our case. As for a second disc for backup: Putting a second disc into the system for backup is not a real backup since if you have a nice failure (powersupply problems e.g.) then both discs might die; and if some app goes wild and starts to delete things a second disc in the system is liable to be affected. For a really sensible backup you want to have two sets of backups (in case one goes broken when needed) and you want to keep them in different places since otherwise they're liable to be worthless in case of theft, fire, ...
mrchocky 1/3/05 9:47AM	Where has this idea about using FAT32 et al come from? No one's suggested using it - it's just a red herring that's been magicked out of nothing in the same way that people suddenly assume the worst with a Windows registry system or a Windows DLL system when we talk about packaging and shared libraries respectively. There are many many types of filing systems out there - some of the Linux ones may be more suitable (but there may be licensing considerations) - it will depend upon the ease to which RISC OS specific code can be written or ported, and how well the filesystem can store additional information required by RISC OS. In other words, please don't jump to any conclusions. *In reply to hzn:* PCs, in any sense of the word, don't know very much about partitioning at all (there may or may not be limited understanding in the BIOS). The specific layout of the disc is understood by the bootloader and/or the OS being run. Any restrictions are those imposed by the partitioning system or the abilities of the OS being run.
jymbob 1/3/05 10:01AM	*In reply to eddie:* It's my Iyo mav's talking about, and by the look of the serial numbers, it appears the first few digits are the model number (IYO11, IYO12, etc) so all the rest would logically be the number of that unit. My question would be: is there a 1 for each model, or do the numbers increment regardless of the model?
johnpettigrew 1/3/05 10:10AM	*In reply to Simo:* FAT32 is widely readable but a horrible format - munged long filenames and fragmentation are just the best-known of its problems. The advantage of ext3 is that it is widely readable; as I said, it's usable on Windows. Personally, I use reiserfs on my linux box, mainly because it's the default FS used by SUSE (the distro I use), but there are no good Windows tools for this format that I know of. *In reply to mrchocky:* I think I was the one who brought up FAT32/NTFS, although I was extrapolating from the sentence "The solution here could be to format hard disc in a standard Windows recognised format" in the original article. I don't think that was an unreasonable extrapolation, assuming that this was a report on what was said rather than an author's interpolation! The reason I suggested a linux format is simply that they are well documented and tested. There are, of course, other options but using an existing format (if it was suitable, of course) would be the easiest.
Use the forum for more comments on this article