Start fresh or use what is currently installed??
Jonah (3312) 9 posts |
Hi can anyone help me out? I just bought a RiscPC 700 with StrongARM and I’m pretty excited as it is machine I’ve wanted for a while as an Acorn and British computer fan. Now it seems to work pretty well but has loads of stuff already installed and could do with a clean up, as well as having all sorts of old documents and apps on there and the old user’s name and named Printers and network devices etc etc!! It is running RiscOS 4.02. My dilemma is whether I should do a fresh install on the machine and completely wipe it – maybe even stick an SSD hard drive in instead of the old one that is already in… But although I’ve used RiscOS on raspberry PI, I’ve not installed it fresh before on an actual classic Acorn machine. I’m not sure how I go about swapping or flashing the rom chips or doing the software install for RiscOS 5. Should I stick with what is already installed, do a clean install of OS4 or do a clean install of OS5?? Any best setups for RiscPC StrongARMs?? Thanks for any help or advice! |
Andrew Rawnsley (492) 1445 posts |
I’d strongly advise sticking with 4.02 on a RiscPC – the big appeal of the older machines is software/hardware compatibility, and 26bit RISC OS 4 is pretty much the best choice for that. You can still run 32bit software with a modern SharedClibrary – available either here or downloading “System Resources” on !Store. The trickier question is “clean” or “dirty”. Ideally you’d want a RISC OS 4.02 install CD – did it happen to come with that? That would have the fresh 4.02 !Boot to build up from. If you can live without some of the OS4 bundled apps, you can download the clean disc image from http://www.riscos.com/ftp_space/400/index.htm The downside is that you’ll lose any extra software that might have been “bundled”, although technically if you don’t have original media…. OK not going there ;) If you do go the clean route, I’d make sure to download !Store from http://www.plingstore.org.uk/ and also grab PackMan too – https://sites.google.com/site/alansriscosstuff/packman If you don’t have internet access (no network card) then probably stick with what you have, as updating the clean image (and getting the clean !Boot onto it) will be a pain in the proverbial. |
Willard Goosey (5119) 257 posts |
IMHO the first thing you should do is make a full backup. Then figure out rather to clean or start fresh. |
John Sandgrounder (1650) 574 posts |
Some good comments on here already.
With RISC OS, it is fairly easy to have both. You don’t need to wipe the exising disc. Just build a new a new one also using 4.02 on an SSD. |
Rick Murray (539) 13840 posts |
Do you have a CD writer? If so, you could use FCFS to take an image of the existing drive (onto another) and dump the data into a CD. |
Steffen Huber (91) 1953 posts |
Which will unfortunately fail whenever FCFS encounters “new filecore” aka E+ format. Not sure if it is good for plain “save that disc as image” for backup purposes only without trying to open it with FCFS. I think the option “do not image free space” or whatever it was called could produce an inconsistent image for E+. |
Rick Murray (539) 13840 posts |
Oh. Okay. I wasn’t aware that it hasn’t ever been updated for E+.
Well if course you try it. A backup that hasn’t been tested might as well be a random collection of bytes taking space on the storage media. 1 Mostly a rhetorical question, as I’m aware of the occasional story of “Instagram” (or some other service) erased my account (by accident) and all my irreplaceable photos are now GONE! NOOOOO!" as if there is absolutely nothing wrong with using somebody else’s free service as one’s only storage of important photos and stuff. Which is a whole magnitude of stupid worse than failing to test the integrity of local backups. |
Steffen Huber (91) 1953 posts |
Since you cannot open it with FCFS (if it is E+), you need to restore it to a different device to test the backup. That was my point really (without mentioning the restore step of course!). |
John Williams (567) 768 posts |
I don’t know if this is any help, but I used to use PackDir (26 bit only) to make my backups on the old RPC. My biggest/latest would easily fit on a CD. An advantage is that, although PackDir has not been 32-bitted, its files can be read by SparkFS on the Iyonix and the RPi under RO 5, thus this is a future-proof solution at the present. I can still read these old back-ups on my RPi from the USB HD, and quite often do so to retrieve old files. |
Steve Pampling (1551) 8170 posts |
I’ve never used PackDir for writing files, but I’ve been a SparkFS user for an awfully long time. The platform independent solution is to TAR into one file and then Zip it (InfoZip or SparkFS) One comment, using a converter to pop in an SSD on a RO4.02 build might be awkward unless the fix for IDE interface issues is in the PlingSystem download and thus available for older OS releases – I think it may well be. |
Rick Murray (539) 13840 posts |
Tar doesn’t preserve filetype information. I wrote a program a long time ago for a friend that made this mistake, it could open tar files and would try to restore filetypes by examining the file to look for certain headers and/or unique bytes/words in specific places. As for FCFS on modern machines, don’t. It, sadly, stiffs the machine when run under Aemulor. But it does work on an emulator, from there one can just push files across the network to the real hardware box. |
Steffen Huber (91) 1953 posts |
PackDir is indeed dramatically quicker than SparkFS. However, compared to SparkFS ZIP Deflate, it has a much worse compression rate. One thing to keep in mind is cross-platform compatibility. For ZIP, it is dead easy (for a developer) to reconstruct the RISC OS specific info from the extra fields on any platform. Decompressing a proprietary format like PackDir’s is also possible, but a lot more complicated. Reminds me that I should really release my Java-based ArchiveViewer application, which can decompress ZIP as well as (thanks to James Woodcock’s efforts) Spark/ArcFS, CFS, PackDir and Squash. I think there was still a problem in the frontend with some strange password-protected Spark files. |
Chris Mahoney (1684) 2165 posts |
May I just point out that ROOL’s source archives are Tar and include typed files? :) |
Rick Murray (539) 13840 posts |
Is that using the comma-type-suffix convention? ;-) There doesn’t appear to be any legitimate place in the standard tar headers in which to store RISC OS specific information… https://www.mkssoftware.com/docs/man4/tar.4.asp …and certainly setting filetypes on tar using SparkFS, it’s forgotten the next time the archive is opened because it’s not a legitimate extension to the format. |
Chris Mahoney (1684) 2165 posts |
I have no idea; I just download them and let UnTarBZ2 do its thing :) |
Steve Pampling (1551) 8170 posts |
Yes. I stand corrected on that score. |
Jonah (3312) 9 posts |
Hi thanks for the replies everyone. I think the “do both” advice is good, if I just buy a cheap SSD drive and then take out the old hard drive so I can put it back if any issues. Currently I’ve also been struggling with my internet setup on the machine. Everything seems to be installed and I’ve got browsers and Network configuration all looks good with TCP/IP ticked. Interfaces is my I-cubed EtherLAN600. Routing has my gateway correctly. Host names has correct name servers. All the settings look good. I can ping other machines on my network, and ping domain names with no packet loss. I can even use the telnet client to connect to BBSes etc. But whenever I open Oregano or any of the other browsers on the system like Oregano, Fresco or !Browse (there’s a load of stuff already installed on this machine) all of them timeout and give a “The requested URL could not be retrieved” type of message… It’s pretty weird that all my FTP, telnet and ping stuff works great but I can’t browse the web. None of this may matter though if I just start fresh on a new hard disk?? It would just be nice to confirm I don’t have any hardware issues. Does anyone know if all SSD drives work ok, or if I need certain adapters etc? Thanks again for the help. |
Steve Pampling (1551) 8170 posts |
If you can ping www.google.co.uk that suggests that your network setup is good. So:
Oregano I think used its own resolver, Fresco I used the ANT stack, Browse I don’t recall.
There’s no SSD capability in the RPC (it’s IDE era) so you need an adapter. |
John Sandgrounder (1650) 574 posts |
Not strictly true. If you are prepared to throw enough money at the problem. But I was thinking of an adapter – although not the one actually needed. [red face]. And, of course, real 3.5inch IDE HDD drives are still available at similar prices to SSDs. |
Jonah (3312) 9 posts |
Hi thanks for all the help. I noticed you can download a RiscOS 5.24 rom – am I able to remove my current rom chips and reflash 5.24 to the roms? Or to be safer can I buy some blank rom chips and flash those so I can swap them out? I say a post saying I’d need 2x M27C160-50F1 – but would these flash with a standard TL866II Plus Universal EPROM USB FLASH Programmer (like this: https://www.ebay.co.uk/itm/High-Speed-TL866II-Plus-Programmer-USB-EPROM-EEPROM-FLASH-BIOS-AVR-PIC/181923646528) or do I need to erase it with a UV and then use a special flasher? Thanks for any help |
Chris Evans (457) 1614 posts |
The TL866II has a 40pin ZIF socket RISC OS 5 ROMS are 42pin! |
Steve Pampling (1551) 8170 posts |
Given that large amounts of the software that came with the machine would be 26 bit wouldn’t trying RO5.2x as a softload be a better first step? If that turns out not to be a nice outcome then remain with RO4.02 and buy a Pi. The Pi could be housed in the RPC case and used for newer software while using the original RPC for older code. |
Bryan Hogan (339) 592 posts |
Can I refer you back to the advice in the first reply – do NOT put RO5 on a RiscPC unless you really know what you are doing. It will be incompatible with most of the software on there, and probably with your network card too! |
Steve Pampling (1551) 8170 posts |
True, most network cards are flashed with code that isnt 32 bit safe (although I think the new code is backward compatible) |