Stable build for Pi

Now that the technical issues preventing this have been solved (but have not yet been implemented due to resource constraints) would a bounty provide an opportunity to fund this task?

I am thinking of Pi model 1 and 2 as well as the Pi Zero. The Pi model 3 will have to wait as those problems have not yet been solved.

Good idea.

I have found the 17-Feb-2015 (5.21) build to be very stable on both Pi 1 and Pi 2.

I use it for shared filestore, Webjames and another HTTP Server written in BASIC on a higher port number. (and, occasionally, !Netsurf). And, of course, my own 25 year old BASIC software.

The only times it crashes are when I do something silly with my software. Otherwise, it runs forever and ever.

Interesting. I’m running 5.21 as well, but it says 8th July 2013. I’ve had it since February 2014. It does seem very robust.

With my latest batch of SD cards from ROOL, i ordered them from a page linked as RISC OS Pi official release on SD card.

I was expecting them to come with an even numbered build, as I thought odd numbered builds (like 5.21) were used only for development and was surprised when the build announced itself as 5.21 (17-Feb-15), as I thought I had paid for an official release – which I thought would be even numbered (I was expecting 5.22).

However, the 17-Feb-15 SD cards all seem to be stable, so I was not too bothered.

It all depends on how you define “stable”. I’m pretty sure that Chris was using the definition of “the developer [ROOL] is happy with it”, as opposed to “as a customer, I haven’t had any issues”. Your usage pattern may not match official testing, and vice versa. Personally, I prefer not to use RC14 as it sometimes suffers from stuck keys; an issue that was thankfully fixed at some point last year.

I am confused. How does “RC14” fit into the 5.xx version numbers?

I agree that the term “stable” is open to defnition. I can rely on the fact that my 17-Feb-15 build will not crash (with what I am doing with it), but I can also rely on the fact that the shift key will get stuck occaisionally!

If I order another *Official Release" SD card today will the stuck shift key issue now be fixed – thus making the new SD Card different from the last Official Release I purchased?

What date will be on today’s SD cards?

All SD cards sold in the past year or so will be RC14, which is the 14th “candidate” release of RISC OS for the Pi (currently 5.21, 17-Feb-15). Previous candidates have been earlier builds of 5.21 or even 5.19.

If you buy a card today then it’ll contain the same version as your existing card, which in turn is the same as the downloadable image.

I assumed the stuck keys I experienced when I first got my Pi was my keyboard – and further assumed I’d confirmed that when a new (£5) keyboard fixed the problem. The only confusing point is that the “faulty” keyboard does work perfectly with my Mac (replacing a Mac wireless keyboard that suffers from a space bar that delivers two spaces more often than not). (Bangs head on desk.)

Is this in any way related to the fact that the mouse on my Pi works fine if plugged directly into one of the USB sockets on the Pi, but is very erratic if plugged into the hub that’s in the other socket? (Whereas the keyboard is perfectly happy going via the hub. As are a hard drive and a memory stick.)

There seem to be two causes for stuck keys. One of them was power draw, so some keyboards would be less affected than others. The issue occurred a lot less when I upgraded to a better power supply. The other problem was in RISC OS itself, and I can’t find the thread now, but from memory a fix was committed about six months ago. I haven’t had a single stuck key since using a fixed ROM.

Edit: Found it.

The version of the USB driver that RISC OS currently uses is based on this version of the Linux driver from April 2014. The Linux driver has been changed quite a bit since then (it’s now using the fiq_fsm USB driver), and now that that’s stabilised one of the things on my todo list is to try updating RISC OS to use it.

Followed by:

Last night I checked in my port of the ‘FIQ FSM’ USB driver, so it should be available in today’s ROM. So if anyone who’s been having trouble with stuck keys could give it a go and report back with their findings then that would be great.

Both quotes are from this thread.

Chris: In the thread you quote you will see that “Last night I checked in my port of the ‘FIQ FSM’ USB driver” was from Jeffrey on the 8th of October and that on the 9th October Colin (Who has done quite a bit of work on USB) mentions his changes (not in the FIQ area but key repeat related) weren’t in the sources yet as: “Jeffreys changes have broken them. I’ll have to see if I can work around his changes.” as far as I’m aware they have yet to be added to the build sources.
Colin: Have I missed your key repeat fixes going into the sources?