Using 5.14

Ok, flashed 5.14 ok. I did find that my mouse had to be unplugged, and replugged in to work – even after power cycling the machine – very strange.

Apart from that, I used UnTarBZ2 to try and unpack the boot stuff. I didn’t enter/drag a destination, which ended up creating the comedy $.ofla..ofla.boot… name (well, ofla as those funky top-bit set chars).

This tells me two things -

1. The app is missing a basic NULL pointer check. 2. RISC OS still urgently needs application zero page read protection. This is probably the single biggest waste of developer time – programs continue on with bogus values instead of immediately failing, leading to all kinds of headaches. The only thing this really breaks is ShareFS, and that should be easy to fix. (e.g. with Adrian’s prot1k).

Other than that, looks good ;-)

He he :) The app doesn’t itself do the work, just some ports of tools like bunzip2 and tar so they will be the ones crashing. However, I think the wrapper Obey file should check for missing params and fail gracefully though. Ta for the error report.

The app doesn’t itself do the work, just some ports of tools like bunzip2 and tar so they will be the ones crashing

I think you’ll find it’s your app passing the ofla filename to them; the tools are just doing what they’re told. There wouldn’t be any meaningful way to pass a NULL to them if it’s just command line based – at worst, they’d complain about a missing parameter or something.

Indeed, my fault. I’ve fixed the version on our web site.

I’ve found a small cosmetic bug when using F12 to get to the command line – one or two rows of pixels from higher up the previous line of text (or graphics if it’s the first line) are repeated below the last line of text after pressing Return. Repeats for each new line, changing to match what ws inthe last line. Only happens with Geminus active. Using flashed 5.14 on Iyonix at 1680×1050 x 16M on an FX5200 card.

Someone should report that to the Geminus author.