RISC OS Desktop Modernisation project

Paolo, just done the equivalent of slipping a note under the door and running away – but you know where I live (so to speak) as the email is monitored whenever I’m awake.

Steve (or someone else) if you would like to take over the admin on the RISC OS Community github that would be awesome

More like a miracle, probably take me forever to figure out github and such if I tried.
I learned today of a couple of project items on top of the existing stuff that are probably going to leave me looking for relaxation away from a screen.

1) What do you think about adopting !ConfiX as default configuration tool for the whole project?

I’ve been in favour of Confix use for many years. It might encourage people to port a few items with no GUI if they realise how easy it could be to do the command switches if they use Confix.

The whole RISC OS experience would be greatly improved by making 3rd party upgrades to functionality the default. But then we get into the “proper stone arrowheads were so much better than these newfangled bronze ones” problem.

Well, depending what sort of stone they were… 8~)

The whole RISC OS experience would be greatly improved by making 3rd party upgrades to functionality the default.

While this makes sense, we run into two potential issues. The first is the licencing, and is it compatible with RISC OS (which means we can discount anything GPL). The second is that one of the strengths of RISC OS is that one can choose whether or not to have these enhancements. It isn’t a forced decision, it’s yours.

What would greatly help is a reliable and up to date source of where to find things, and pointers to that in the distributions (and on this site).

Sometimes tracking down the right version of something to work on newer hardware (WabJames and Samba to mention two) is more difficult than it should be. I’d include Zap in that list, but since I’ve stuck it into Store it ought to be simpler to find. Well, that’s the theory. ;-)

The “other side” has a products database that has so many broken links it isn’t funny. But, then, that list predates my coming to France, and I’ve been here nearly twenty years…

The whole RISC OS experience would be greatly improved by making 3rd party upgrades to functionality the default.

To suit different tastes, you need two things: a base package, and well-designed interchangeable and isolated add-ons. An integration of the base package and severall add-ons would then be a distribution like RISC OS Direct. However, you need a dedicated bunch of people to keep things up-to-date – see the RISC OS Direct disaster where there wasn’t a Pi 4 compatible version for several months. If you want to provide something beginner-friendly, you cannot afford a cock-up like that.

Bundling things with the base OS that is developed under the auspices of RISC OS Open – I don’t think this is a good way forward. I would even remove applications like Draw, Paint, ChangeFSI and Maestro from the base disc image.

The base package could be fairly minimal – the bare OS and a package manager. This also makes licensing mostly a non-issue, since you delegate the “integration” to the user. So no more GPL headaches.

If you want to provide something beginner-friendly, you cannot afford a cock-up like that.

Are you volunteering to help? Because with limited numbers of developers and developer time, it is more a case of “it’s ready when it’s ready”.

The base package could be fairly minimal

Which seems to be the opposite of everything else these days. Linux distributions, Windows, iOS, Android, they all come with certain functionality built in.
We need Paint and Maestro to gloss over the long list of things that the OS doesn’t do out of the box (such as pull AAC audio from a WiFi access point and push it to a pair of Bluetooth headphones ¹).

So no more GPL headaches.

Right up until somebody suggests ditching the old DOSFS and replacing it with Fat32FS. Which would be good given that Fat32FS works, DOSFS not so much. But, um… Yeah. Oops.

¹ That’s not intended to show limitations in the OS, although it does. ;-) It just happens to be what I’m doing with my phone right now. And once I’ve finished writing this, I’m going to waste some of my life with Netflix…

@ Steffen

To suit different tastes, you need two things: a base package, and well-designed interchangeable and isolated add-ons.

That’s exactly the architecture of the (temporarily called) DME.

The core is minimal and handles the configuration and activation of a set of add-ons. Such add-ons are being designed and built either as regular (but really small) Apps or as TaskModules (where absolutely required).

add-ons interface or control certain behaviours of the WIMP or the Filer and for phase 1 that’s already plenty of work.

see the RISC OS Direct disaster where there wasn’t a Pi 4 compatible version for several months. If you want to provide something beginner-friendly, you cannot afford a cock-up like that.

Steffen if you want to help us, that would be great. Also given the issue you’re mentioning we are trying to put effort on using BBC BASIC where possible and sticking to ROOL guide lines and Toolbox and Style. To avoid as much issues as possible.

@ Rick

While this makes sense, we run into two potential issues. The first is the licencing, and is it compatible with RISC OS (which means we can discount anything GPL). The second is that one of the strengths of RISC OS is that one can choose whether or not to have these enhancements. It isn’t a forced decision, it’s yours.

Exactly and with this effort I am also contacting people to get written permission to use their previous code and make sure that all the code we are going to use, not only will have a ROOL compatible license (yes I am also asking if we can change the license with Apache 2.0), but we are going to review the sources and adapt them to work on RO 5 (where needed) and still support older RISC OS.

What would greatly help is a reliable and up to date source of where to find things, and pointers to that in the distributions (and on this site).

Absolutely true. We are going to stick to RiscPkg initially to deliver the code in an easy to upgrade way. The package will also be available for download and I am looking at adding an installer to make sure everything gets installed correctly and in the right place even without RiscPkg.

Right now the idea is to provide a repository of all the add-ons and themes and in a single place (maybe inside of the usual !Boot.Resources maybe somewhere else). Questions will be asked to all of the community on here in the General forum, so up for vote from everyone, hopefully this is a good thing :)

@ everyone

Sorry for have been quiet for a while, just got my hands full at work again, but we are making progresses every weekends and again if anyone wants to help please feel free to join, info and details will soon be available, we are working out some initial guidelines to join the effort and they will be submitted for everyone’s review soon if that’s something you would like to do :)

Also to be absolutely clear on github we have created a RISC OS community organisation, so everyone can join to help even who cannot provide code and again if you have a tool that would like to submit, please get in contact. Here is ok, on my blog is ok, email, whatever you like.

Project is here:

https://github.com/orgs/RISC-OS-Community/projects/1

Presentation is on the link on the first page and has been reviewed by Steve P. so my English there is not as horrible as usual! :D

so my English there is not as horrible as usual!

Mine is :)

Sorry to derail your thread, Paolo…

If you want to provide something beginner-friendly, you cannot afford a cock-up like that.

Are you volunteering to help? Because with limited numbers of developers and developer time, it is more a case of “it’s ready when it’s ready”.

Keeping RISC OS Direct up-to-date is not a job that needs developer time. Putting up a warning in big red letters near the download link to make it absolutely clear that this distribution does only work with the Raspberry Pi models up to 3 and unfortunately not with the Pi 4 and the Pi 400 takes a few seconds.

I involuntarily volunteered to explain the Pi compatibility problem to some newcomers, and you can be absolutely sure that they were not happy. The complete lack of a version history and indication of what exactly is the state of that rather large download (notice the lack of information about the size?)

Apart from that, the RISC OS Developments website is completely out of date (notice the “Events & News” page), so whatever volunteer time they get (which seems unlikely, because they don’t even ask for volunteers) should first be put into that problem.

The base package could be fairly minimal

Which seems to be the opposite of everything else these days. Linux distributions, Windows, iOS, Android, they all come with certain functionality built in.

That’s because they are full distributions and not the base package to which I am referring to. I thought I made that clear in the part that you have snipped.

We need Paint and Maestro to gloss over the long list of things that the OS doesn’t do out of the box (such as pull AAC audio from a WiFi access point and push it to a pair of Bluetooth headphones 1).

The distribution I am referring to would still (probably – it is of course the choice of the packager) contain Paint and Maestro, so I am not sure what exactly your point is.

So no more GPL headaches.

Right up until somebody suggests ditching the old DOSFS and replacing it with Fat32FS. Which would be good given that Fat32FS works, DOSFS not so much. But, um… Yeah. Oops.

As long as Fat32FS is a downloadable package that augments the base distribution, this is not a problem at all. The GPL is fine as long as you don’t distribute GPLed software as part of your own package (and even then, it might be OK – I would argue that including Fat32FS in the HardDisc4 download would be OK from a license point of view).

does only work with the Raspberry Pi models up to 3 and unfortunately not with the Pi 4 and the Pi 400 takes a few seconds.

True.

The complete lack of a version history

Also true. One of the things that annoys the hell out of me with Android apps is when the version history says something like “Bugfixes and improvements”. Yeah, WHAT? Every app has “quirks”, it would be nice to know if my favourite bug quirk has been fixed, or not.
But, alas, few people bother (especially Google) which makes the entire “what’s changed” thing seem rather pointless for the majority of apps.

So, yes, if one is going to make a BIG download, it would be better to know what “Now compatible with the Pi 400!” rather than “some shiny new green icons” before downloading.

the RISC OS Developments website is completely out of date (notice the “Events & News” page),

<clicky clicky>

THIS YEAR
RISC OS SOUTHWEST SHOW 2020

That’s… actually pretty good for a RISC OS website.

Getting people to update their bloody sites is an uphill battle. There is still the mentality of the ’90s where anything interesting happens at “a show” and the site is a sort of ginger cousin that nobody wants to play with.
Bzzt! WRONG!
These days, the site is your primary contact with potential customers, newbies, and other interested people. Having a site that headlines an event from fourteen months ago is not a good look. Yes, I get it, Coronavirus/Covid/Covid-19/SARS-Cov-2/TheBloodyPlague came and shook everything up. But, has nothing of interest happened in the RISC OS world since then? Nothing at all?

[to be fair, the most recent thing mentioned is the London show in October, which is a few entries down as it’s a chronological jumble; so, only about six months out of date… that’s really good for a RISC OS site… yes, I’m taking the piss…]

should first be put into that problem.

Will “lame ass barely updated websites” win the broken cog award this year?

I’ll repeat it again – websites are the first contact people will have. Use them. Don’t treat them as an afterthought.

That’s because they are full distributions and not the base package to which I am referring to.

The point is, people go for the full versions and not a base install. Because getting this and that and the other and setting it all up is a monumental pain in the backside. Especially if you don’t really know or understand the system (you’ll notice an absence of fancy installers in RISC OS, except for a few commercial products that use it for “copy protection”).

The point is, people go for the full versions and not a base install.

It would be nice to be able to do that with Windows too. The last time I had to do full install (after a disc failure) it took a couple of hours to install Windows, and 2 days to install all the other stuff (including 3rd party drivers). (And that was with most of the downloads already in place from the previous time, on a non-failed partition on the original disc.)

The real question is “what is a full version?”.

These days, the site is your primary contact with potential customers, newbies, and other interested people. Having a site that headlines an event from fourteen months ago is not a good look. Yes, I get it, Coronavirus/Covid/Covid-19/SARS-Cov-2/TheBloodyPlague came and shook everything up.

Apparently it’s not shaken everything up.
Despite change in some quarters¹ some things seem to plod along just the same: I could swear I was mumbling/complaining/whatever about hard to find/non-existent info about products on RO web sites for most (if not all) of the last ten years.
“No time to update”, “lots of effort preparing for XYZ show” were among the excuses.
So, stuck inside with loads of time on hands, and we get? No change.

I’ll repeat it again – websites are the first contact people will have. Use them. Don’t treat them as an afterthought.

First contact? The web is currently the only contact.
Get used to it, although you’re probably happy to have been in the same position as any other RO user with regard to show attendance Rick.
Work commitments permitting, you (like others) could probably attend every ROUGOL meeting without leaving the house.
Web sites are the everyday part, the Zoom meets and shows are the extra interest.

¹ Some people restarted work (again) last Monday, some of us never stopped² although I have only been in the office 4 times in the last 7 months and one of those was scooping up the contents of my desk³ drawers.

Sorry to derail your thread, Paolo…

Oh I see you guys were talking about something else… I thought the comments were about the project in the topic. Then sorry Steffen I misunderstood.

I think !ConfiX can be the standard for RISC OS applications to create and manage tabs.

One bug report of the current RISC OS Theme manager. It allows only very short folder/theme names.
If you add folders with longer names it simply refuse to work.

Overall sounds like what we can already do, just made easier.

Yup that’s the point of it:

• Easy to use for everyone, included who has never seen RISC OS before
• Single point of distribution of all the included Utilities (so no need to go around the web to get bits and pieces)
• Consistent Theming protocol (instead of “broken into unrelated pieces” as it is now)
• All added features are OPTIONAL, this means that users who like the purest and oldest form of RISC OS Desktop can have it their way, while users that wants a more modern experience can have it their way too (and everything in between!)
• All extensions and configurations are file based, so hardware vendors like RISCOSBits, RComp, Elasar, CJE etc. can decide their default desktop behaviour, look and feel and pre-install their config on their SD images.
• Extension of the Desktop experience through the addition of extra tools (so no WindowManager modifications which also means it’s compatible with all the existing software, if anything the existing software may not use new features, but it will still work fine)
• Adding new tools to RISC OS Desktop like:
  • !FirstBoot an initial system configuration tool (gets executed only at the very first boot, OR if the OS has not yet done first configuration, OR if an OS update has requested to re-configure a portion of the OS). It supports also partial reconfigurations (these can be needed post updates, after reboot).
  • !OSUpdates (when done) will enable easy OS updates (at this time I am working on ways to make this work, so very very early stages)

Which would be good for new users at least.

Yup, and if !FisrtBoot will come as I wish it would, it would be very useful for old users too, I am planning to add code to auto configure email clients like Messenger Pro or Pluto (if they are present on the SD Image). !FirstBoot works a bit like Apple Mac configuration tool. It’s a “wizard” that allows initial configuration based on what an hardware vendor wants the user to be able to configure initially.

To the question of a name what about ‘Desktop Configuration Plus’. As it is primarily a matter of adding existing freeware components, and simplifying the configuration of the whole this seems a sensible name to me.

Thanks for the vote. Yes it’s a combination of features and new Theming. One of the reasons to call it “Desktop”-something is purely for usability, a lot of Linux Distros, macOS and few other systems use the term Desktop for such type of configuration in their System Preferences, so it may help to make it a bit more user-friendly and/or intuitive.

Why do your screen shots show much retro looking setup? You advertize as modernization, yet the flat look went out in the late 1980s to early 1990s.

I have to correct you here. The flat look has been back and for quite a while now: Windows 10, macOS all latest versions, Linux KDE 5.18,.20,.22 and Gnome Desktop latests. So no your statement is wrong I am afraid, please google for these Desktop screenshot before commenting back, thanks.

I must dissagree with the archetecutre you propose. That is me, see my statements above.

Not a problem at all, I design code as it pleases me, RISC OS for me it’s only fun. Totally respect all the disagreements :)
However, please do not confuse the term “Architecture” with the term “Look” because the comment you are referring to, in your statement, talks about look, not about Architecture. So, I think you meant you do not agree on what is considered a modern desktop look. In which case, I would suggest: GOTO my previous response (reins and repeat until you have looked at all the relatively new desktops, they are all flat), thanks.

RISC OS Select tried that. I think most people went out of there way to dissable it. I think it was select anyway, one of the RISC OS distros for sure.

That’s the whole point of my project, allow people to configure the look and feel as it pleases them. I am not forcing anyone to do anything. Plugins can also be replaced and, where there is the possibility there will be already a choice of different paths.

There are also some freeware utilities that do this.

Again this is the point of the DME (Desktop Modernisation Engine), to use already made software where possible (as long as the original author allow me to and responds to my email! loool).

What is wrong with what we have?

I am sure Rick and others could fill up books of comments lol. RISC OS has a lot of idiosyncrasies, one of them is that portions of the kernel defines short cuts directly instead of using a dedicated module for this. The results (beside making the sources more complex) is that behaviours can be inconsistent. But, again, on DME everything is optional, so no one is forced to follow any rule.

Sounds like Justins idea of a pinboard server for sillies the way you describe it. Also sounds primarily of use to pinboard sillies.

Not sure what you mean with the term “Sillies”. So do not understand what you’re trying to say or imply, sorry.

As a general answer:
1st I am profoundly grateful to all Justin has done and to all the info he has shared with us.
2nd Gadgets can help to improve desktop experience AND we actually have quite a few of them.
3rd I think that the new pinboard being developed by ROD is actually very useful and very welcome.

So, I do not see any “sillies” here, except from some too superficial comments maybe? ;)

@ Stefan

One bug report of the current RISC OS Theme manager. It allows only very short folder/theme names.
If you add folders with longer names it simply refuse to work.

That sounds more like “compatibility” with the past and short names maybe?

On the DME there is no forced file name length or directory name length, however if a Theme creator uses long names then his/her theme may not be usable on old RISC OS versions that do not support long file names.

The flat look has been back and for quite a while now

Oh, god, yes. Apple threw out their skuomorphism ¹ in the transition of iOS6 to iOS7 and made everything boring and flat. And then had to release an update that had an option to put a darker bit behind buttons so people knew what bit of the screen they were supposed to poke.
Then Google jumped on the bandwagon. And so did, in time, everybody else. I remember one of the more complimentary things I said about iOS7 was that it reminded me of GEM.

If I sort myself out a newer phone (I was supposed to back in February, my contract has technically expired, but that would mean interacting with potential-zombies) I’m sure it’ll be running a version of Android that will look and feel different to Samsung’s take on Android 8.
I do wish everybody would stop mucking with the UI “because they can” (that’s the only excuse that makes sense).

What is wrong with what we have?

I am sure Rick and others could fill up books of comments lol.

So do not understand what you’re trying to say or imply, sorry.

Nothing nice, I would imagine. A “silly” is an app that is, well, silly. Briefly amusing but ultimately pointless.
Two I recall from the Archimedes days:

• A window with two big eyes in it. The centres of the eyes moved, so it was always looking at the mouse pointer.
• A sprite of a yellow rubber duck that drifted after the mouse pointer.

I wrote one of my own a long time ago – a set of Christmas lights across the top of the screen (and since I’m me, it had a fuse bulb and also randomly blown bulbs (did I ever release that version?)).
This is, of course, a giant anachronism these days. Every blinkenlights set I’ve used since at least ten years has used LED lights and either run off batteries or a power brick. The orange lights I use on the pine tree out front are 24V and all LED based.

That’s what “sillies” are, anyway.

¹ Fancy word meaning “nice wood effect backdrops and buttons that look like buttons”.

That sounds more like “compatibility” with the past and short names maybe?

No, it’s a real problem with Theme Manager at RISC OS 5.28.

To reproduce the problem:

• Go to !Boot.Resources.!ThemeDefs.Themes and copy the Sovereign directory to a directory called “DeliberatelyLongThemeName”.
• Go to Configuration > Themes – you will notice that the option “Load extra theme resources on startup” is now greyed out, preventing you from selecting a different theme.
• To restore normality, delete the “DeliberatelyLongThemeName” directory.

@ Rick

That’s what “sillies” are, anyway.

Thanks for the explanation! So in theory one could do those too, but the ones targeted for the gadgets are more “useful” maybe:

• Show clock rate, time memory available (yes live on the desktop instead of using the icon bar!)
• Show RSS feeds if one wants
• (future) Show Tweets one is interested to and have a click option to open a browser and get there
• Notifications (from the notification framework if someone has it enabled)

And so on… but if one wants a couple of eyes following the mouse ala early X11, it’s up to you sure XD

I wrote one of my own a long time ago – a set of Christmas lights across the top of the screen (and since I’m me, it had a fuse bulb and also randomly blown bulbs (did I ever release that version?)).
This is, of course, a giant anachronism these days. Every blinkenlights set I’ve used since at least ten years has used LED lights and either run off batteries or a power brick. The orange lights I use on the pine tree out front are 24V and all LED based.

I still have a set with a fuse bulb and a couple of blown ones. It’s a hand-me-down from my parents, and was apparently a promotion that Kentucky Fried Chicken was running at some point (before the name changed to KFC).

Edit: Oops; this isn’t Aldershot. I’ll shut up now :)

bug report moved

And so on… but if one wants a couple of eyes following the mouse ala early X11, it’s up to you sure XD

It was useful on early Sun workstations with the huge monochrome million pixel displays, as often it was hard to find the mouse pointer amongst the stippled patterns. If you watched the eyes it was easier to track it down, and it would let you know if the red/green lined laser mouse pad was orientated the right way.

yup, I remember… ermmm I should say I still have some of these things up and running here! XD

I have a ton of the old things still in working order, when I have time I enjoy fixing old machines and still run plenty of old things from 8bit to old OSes included OS/2, XENIX, early Linux with CDE, RISC iX (of course!), bunch of DOSes and early Windows, early macOS 6,7 and 8, various Amiga OS, Atari TOS and more.

It’s just fun and takes my mind away from work, reduce stress and give me good memories :)