ARM BASIC: SOUND extension?

Samba v2 and v3, Ipv6, Firewall,

IPv6 is one of the items in the stack rebuild, but that’s in stage 3 or 4 and the stage 2 stuff is the current bounty development. Gotta build the stack from the bottom up.

Samba – no point building onto the old stack. Supported versions is an interesting one as things are becoming more secure and old versions get deprecated.
Firewall – ditto

Extras? Or did those already get listed earlier?

I still prefer Select RO6 to 5 in mayway, and and fell more at home on 6 then 5, sorry but that life.

The aim is to produce a version that everyone feels at home with¹. Part of the sequence is to identify the things that make people “feel at home”.

Truth be told RO is outdated (notice I had no version number on that.) but that’s possibly part of what makes people feel at home.

¹ For large parts of the feature set it will probably require a little more than “turn on/off” to keep people happy.

Samba – no point building onto the old stack.

Firewall – ditto

Neither of them is affected by the stack.

Really need a new thread for this

Think that would be a good idea.

As to what I’d like to see then I would go for getting the main elements updated as per the bounties first before we add some gloss over the OS.

Without updating some of the core components first then the OS will be like the equivalent of painting the interior of a home that has subsidence and is about to fall down.

Once core elements are updated then we can add some of the gloss and I agree some of the elements of the ROL branch do add things like:

- Image Renderers/Converters built in
- Updates to filer window as per Martin Bazley’s post of 28/10/13
- Updated Boot options including ability to disable/remove entries
- Internet configuration options updated as per RO 4.39/6
- Shutdown options updated so you can get out of an accidental shutdown

In addition better support to talk to other systems with more modern security built in.

I also think Rick’s suggestions are spot on as well so that we can recover easier from errors in Boot or disc errors

As to what I’d like to see then I would go for getting the main elements updated as per the bounties first before we add some gloss over the OS

I think the “gloss over the OS” elements are probably in the realm of what ROOL refer to as small projects. As such almost anyone can have a go.

Updates to filer window as per Martin Bazley’s post of 28/10/13

This Mimemap post? – I can’t find another and definitely not about the Filer. Mind you Mimemap entry maintenance is one of those “where the hell is it and how do I edit it?” situations. Does have a solution – with an add-in

Updated Boot options including ability to disable/remove entries

Probably best done as part of a boot structure revamp

Internet configuration options updated as per RO 4.39/6

Covered in the TCP/IP stack overhaul – specifically “Changes to support configuration of the stack, for example through InetSetup” – whether ROOL would like that as per the ROL work is unknown.

Shutdown options updated so you can get out of an accidental shutdown

Didn’t Rick say something about having done something?
Again, with access to the source you don’t have to hack in an attempt to capture the event before the OS shuts down, you make the checks part of the shutdown.

I can’t find another and definitely not about the Filer

Well that will teach me to actual quote the right person who qouted Martin Bazley and check the source !

So it was Chris Evans on 20/04/16 , quoting Martin Bazley from 28/10/13 who collated a list of missing features

Oh you just gone and made me look at how to do a link in Textile :-)

In case it doesn’t work.

– Double-click renaming.
– Thumbnail display.
– True numerical sort order.
– Display sort order in title bar.
– Filetype menus (customisable) for ‘Set type’.
– Toolbars.
– Per-directory and inherited sort orders.
– New keyboard shortcuts for copy, change display, new dir, run.

So it was Chris Evans on 20/04/16

I wonder if Chris could see his way to funding some of these things?
No matter, some of it is already there now.

Oh you just gone and made me look at how to do a link in Textile

A learning day is a good day, you’re still alive and the brain cells function :)

Some I question:

Double-click renaming.

As opposed to Alt-Click ? or Ctrl-R on a selected file (both in the OS right now, but ^R is a secret squirrel implementation)

Display sort order in title bar.

Slightly puzzled there as the order of the icons is usually more of a clue than a small text item tucked into a title bar, particularly with my eyesight…

Filetype menus (customisable) for ‘Set type’.

That I like, provided the longer filetype lists don’t slow things down.

Toolbars.

Not sure what that would be for. Avoiding context dependent menus?

Per-directory and inherited sort orders.

Useful. Didn’t someone do an add-on?

New keyboard shortcuts for copy, change display, new dir, run

Copy in what context? Filer – logically that would be Ctrl-C I suppose.
Change display? Are we referring to Filer Layout as in Large icons, small icons, and full info? ROOL set Ctrl-F to toggle through the option I prefer Ctrl-L with the “Display” in the menu changed to the more explanatory “Filer Layout”

New dir? Ctrl-D on the ROOL version, I prefer Insert

Run? Ctrl-M (which is Carriage return/Enter) seems sensible to ROOL, and in fact Keypress already captures it – sadly there’s just a ToDo where the action code should be. I hacked in a little something and it doesn’t crash on me.
Mind you if you have the mouse in action to select the file or app you simply double-click so you’re wanting keyboard control of the pointer which is another set of keys to be done.

– Toolbars

One of the nice things about RISC OS is that it doesn’t have such clutter.

As for all of the rest of your suggestions – it’s been done. Filer+ for the original machines, then FilerPro (sold by Warm Silence) for the RiscPC era. Both complete Filer replacements, written by Thomas Olsson – http://www.armpit.dk/files/filerpro/

then FilerPro (sold by Warm Silence) for the RiscPC era. Both complete Filer replacements, written by Thomas Olsson

IIRC, there was never a release version of FilerPro (originally called ProFiler, again IIRC – it was back when FidoNet was the hot thing and I was of fan of FidoMail, again by Thomas Olsson). I know that there was a beta test phase (I was asked to participate because of my involvement with WSS via CDBurn), but I don’t think it was ever sold.

Rick and Steve

The Toolbars are customisable in select – they provide a couple of basic ones, and can be on or off., as for these third party add on all very well if the authors are still around and support/promote the addon, but they would be better in the OS and then can be maintained as part of the OS.

Just becuase you don’t fancy them does not mean others should not have them. The Boot startup task option to suspend and an app being run at start up makes finding a fault easier and saves having to redo them, all again etc.. Simple thing can help boost the OS and make it more friendly/modern.

The Toolbars are customisable in select –

Ah, never seen Select in final incarnation. Neutral on that.

as for these third party add on all very well if the authors are still around and support/promote the addon, but they would be better in the OS and then can be maintained as part of the OS.

I mentioned there having been something that per-directory sort as I vaguely recalled seeing something along those lines. A quick check on the FilerPro stuff mentioned by Rick reveals a 26bit reliant item so testing that to see what’s in it is problematic.

Agreed these things need to be in the OS wherever possible. In some cases though the add-on features are fairly complex – FilerPro for example totally replaced the Filer according to the info I seen

Just becuase you don’t fancy them does not mean others should not have them.

I regard that kind of thing as a prioritisation marker, more interest, more priority.

The Boot startup task option to suspend and an app being run at start up makes finding a fault easier and saves having to redo them, all again etc.. Simple thing can help boost the OS and make it more friendly/modern.

Yes, but the Boot needs more than a little tweak to make it even vaguely like the stuff I’ve seen described in Select. Per-user boot? Hence the suggestion of a wholesale revamp which would include the small aspects like selectable disable on elements.

PS Last posting before the server broke was by…

Agreed these things need to be in the OS wherever possible. […] FilerPro for example totally replaced the Filer […]

Yes, but when the OS is closed source there’s not much choice. Thankfully that’s not the case now, though at the same time we’re suffering from greater access to the source but fewer people who can understand it.

Per-user boot?

RISC OS doesn’t support per-user directories (not even like Econet of old, the /WR permission is essentially useless); so what does the per user boot actually do, other than simply switch one boot script for another (shouldn’t be too hard to stick Choices into a subdirectory by name) so everybody can have customised icons and tasks running? If so, I’d put this into the “visual fluff” category.

Note – Harinezumi didn’t support any of this. Never seen it, don’t know if it’s how I described or something more complex.

Interestingly enough – this doesn’t mention per user booting: http://gerph.org/riscos/ramble/bootmenu.html

Interestingly enough – this doesn’t mention per user booting: http://gerph.org/riscos/ramble/bootmenu.html

http://gerph.org/riscos/ramble/earlyapps.html shows the “WhoAreYou” and “AreaFiler” apps, but I have no idea how much (if any) of this made it into OS 4 or 6.

As for all of the rest of your suggestions

Well not mine but as a result of a survey that Martin compiled the list from of features at that stage that people wanted from RISCOS 4.39/6 in 5.

Ah, never seen Select in final incarnation.

Well as others have said it has more of a gloss about it than 5.xx does but perhaps now doesn’t have the more current core elements as development effectively is dead.

As I said earlier I’m for concentrating on the updates to the core elements at the moment but that doesn’t mean that I do not miss some of the shine that 4.39/6 has in other areas.

I also agree that !Boot is long overdue a major overhaul and at that stage perhaps some of the more polished configuration elements could be put in but for now I’m settling on focusing on the bounties and perhaps the ROD element of a more feature rich browser experience.

In this thread I saw many interesting applications for solving/mitigating some ‘quirks’ that RISC OS currently have.
My experience with Acorns is limited to 32-bit era, but I always thought that Acorn at first let third parties made some solutions to problems/features as technology advanced (ex. DOSFS, NewLook) and successively incorporated it in the OS.
Could someone ‘filter’ the apps in this thread and ask for include these in the current stable OS distribution? if the users find them valuables, their functionality could be incorporated in the main OS (with permission from authors obviously).

Before chocolate, Steve wrote

Probably worked nicely on a 1980s TV and similar.

Indeed. But the corollary is that on a 1980s TV you wouldn’t notice if it wasn’t working very well.

The other thing about all those calculations for the level of influence on a single pixel is the computational load.

Proper font engines don’t do this. They generate AA bitmaps direct.

Since Nemo was saying RO5 was bad

No no. I’m saying RISC OS is bad. IYSWIM. Not a five-specific thing.

Rick empathised

I feel like joining Nemo and banging my head on something solid. Then I realise it’s causing pain to the wrong person.

Some people would disagree about that. In both cases. ;-)

Just think what the RISC OS market would be like if the RiscPC was 32 bit from the outset

It would be exactly the same. Other than a surprisingly small set of compromised APIs (HeapSort et al) the problems have been caused by lack of parameter checking, promoting “programming by coincidence”.

Truth be told RO is outdated

Without wishing to invoke the “some of us are more equal than others” trope, it’s certainly the case that some versions of RO feel more outdated than others, in obvious places.

Doug added

Image Renderers/Converters built in

I’ve been looking into this, and it’s all a bit half-hearted. A new thread is needed for this subject, which I’ll definitely come back to. It is related to clipboard handling, which is also bad everywhere in RO.

Updates to filer window

Hooeee have I got some interesting Filer things in the pipeline. I’m saying nothing.

Shutdown options updated so you can get out of an accidental shutdown

RO4 has a nice shutdown confirmation window (which can even quit the emulator, hurrah) but the protocol still fails when something objects to the shutdown so it can save, and then restarts the shutdown – no further confirmation at all! :-( Needs fixing.

FilerPro

Although it shall always be the case that Other Magazines Are Available¹ the RO5 Filer does need to catch up, and there’s other things you will absolutely want it to do once I show you it doing them.

what does the per user boot actually do, other than simply switch one boot script for another (shouldn’t be too hard to stick Choices into a subdirectory by name) so everybody can have customised icons and tasks running? If so, I’d put this into the “visual fluff” category.

The primary reason is to have per-user choices files. But multi-user RO does assume one has friends that want to use your RO machine. Or indeed any friends at all. :-/

The !Boot structure has evolved over time, but the RO4 multi-user implementation is unnecessarily complicated. It can be very difficult indeed to work out where something is coming from (“I didn’t load that module!”), because in addition to multiple users there’s multiple hardware environments. The resulting !Boot structure is over-deep. Needs flattening somewhat.

However, !Boot is RO’s equivalent of C:\Windows\ and no one ever goes in there if they can help it. So… if it works, it works.

¹ BBC form of words for those of a foreign persuasion

It is related to clipboard handling, which is also bad everywhere in RO.

Maybe we could extend that to “clipboard handling is bad everywhere in every OS (I know of)”.

But the corollary is that on a 1980s TV you wouldn’t notice if it wasn’t working very well.

I was sort of hinting that but I suspect that the effect was actually an improvement on nothing for that particular use case.

No no. I’m saying RISC OS is bad. IYSWIM. Not a five-specific thing.

Ah, clarity and understanding.
It would therefore be rather nice if someone put something different in place.

Rick empathised

I feel like joining Nemo and banging my head on something solid. Then I realise it’s causing pain to the wrong person.

When I said that I mentally compared work and leisure instances of management “help”
I’ve just had a day full of problems that have a human cause – NHSNet causing issues with systems because of naff DNS (again) and claiming different (again)
“It’s your DNS supplying the wrong IP’s” they say. “Here the hell do you think our DNS gets its info from?” say I.
It gets rather old and I shifted to strong words and tone to them rather than even contemplating pain for me.

Some people would disagree about that. In both cases. ;-)

:-P

Other than a surprisingly small set of compromised APIs (HeapSort et al) the problems have been caused by lack of parameter checking, promoting “programming by coincidence”.

Now there’s a coincidence because I said that most of the zero page problems came from programming mistakes that just happened to have not been spotted. I first said it a few years ago and referred back to that in this thread because ROL appear to have spotted the issue and swept it under the carpet.

Without wishing to invoke the “some of us are more equal than others” trope, it’s certainly the case that some versions of RO feel more outdated than others, in obvious places.

The good thing about obvious places is that they are easy to locate and take corrective action. How difficult the corrective action is to implement will vary.

Hooeee have I got some interesting Filer things in the pipeline. I’m saying nothing.

Is it just me or is the Filer doing things that should be Wimp? Once you start into all that Keyboard, Wimp, Filer chain it’s probably tempting to look at the API and replace the lot with something that behaves similarly enough that applications don’t throw a paddy…

the RO5 Filer does need to catch up, and there’s other things you will absolutely want it to do once I show you it doing them.

There’s alpha test volunteers around you know.

The primary reason is to have per-user choices files. But multi-user RO does assume one has friends that want to use your RO machine. Or indeed any friends at all.

Or different use cases that don’t, perhaps, need the baggage inherent in having a non-specialised setup trying to do all possible things at peak efficiency and failing.
Friends vs. Friends that use the same computer, same OS on their own computer, or friends that have an android tablet and don’t give a “great deal”¹ about desktops.

The !Boot structure has evolved over time, but the RO4 multi-user implementation is unnecessarily complicated.

It’s a solution looking for a problem. It could actually be a “camel” and have features that fit various different simpler models.

The resulting !Boot structure is over-deep. Needs flattening somewhat.

However, !Boot is RO’s equivalent of C:\Windows\

I’d say ripping out certain items would help – even MS made moves to slim things down by putting application data in a different location. Bad implementation that loads the profile down, but…

¹ Other comments are available.

I’d say ripping out certain items would help – even MS made moves to slim things down by putting application data in a different location. Bad implementation that loads the profile down, but…

IMHO !Boot should be less monolithic.

A simple !Boot folder, with totally independent !Boot.!System (core utilities linked to the current OS version), !Boot.!Fonts, !Boot.!Internet, !Boot.!Modules, !Boot.!Prefs, !Boot.!Scrap, etc. All coming with their own configuration tool. Double-click on !Boot = open it. Then double-click on the needed subdirectory to get the related parameters. No configure application any more (the general parameters tool could be in !Boot.!System, with parameters stored in CMOS).

A way to ease an OS update, just by updating !Boot.!System (and leave the others untouched if not updated).

nemo:

But the corollary is that on a 1980s TV you wouldn’t notice if it wasn’t working very well.

True. The dot pitch on some of those CRTs was abysmal.

And let’s not talk about when loads of us used AKF12s in 2:4 modes.

No no. I’m saying RISC OS is bad.

I think there shouldn’t be any doubt about your… unhappiness with the “capabilities” (in scare quotes) of FontManager.

Thing is, FontManager is what we’re stuck with.

Rick empathised

I feel like joining Nemo and banging my head on something solid.

That was actually Steve, but I agree so feel free to pretend I said it too. ;-)

It would be exactly the same.

Well, we wouldn’t have the old/new split, we might have more 32 bit compatible software (like Impression)? That’s not to say it wouldn’t break with unaligned loads, ARMv7 tightening, SWP, and all the other changes along the way, but… May have been a healthier continuation?

It’s all just mooting, anyway.

It is related to clipboard handling, which is also bad everywhere in RO.

Yup. In the core of NetSurf, copy-paste works as expected but it has omissions due to how RISC OS itself behaves. I can ^C text up there, ^V down here, but I can’t ^V a URL into the URL icon.
Instead I need to Menu on the URL, export the link as text, and drag the file to the URL icon. Hmmm…

RO4 has a nice shutdown confirmation window

So does my machine. I trap the keypress and give an option. It’s about time Switcher did that for itself…!

and there’s other things you will absolutely want it to do once I show you it doing them

Why do I sense an animated GIF on its way?

But multi-user RO does assume one has friends that want to use your RO machine.

…and don’t say “the hell’s that?” (or words to that effect). Can it do Facebook? (head → table)

Or indeed any friends at all. :-/

Ouch!

!Boot is RO’s equivalent of C:\Windows\ and no one ever goes in there if they can help it.

Nah, nothing is like the mess inside C:\Windows!

Steffen:

Maybe we could extend that to “clipboard handling is bad everywhere in every OS (I know of)”.

I quite like Android’s multiple-level clipboard, but I really don’t like the fact that nudging a handle just a tiny bit can do insane things (that usually involves selecting a thousand pages except the paragraph you wanted). Really, the selection autoscroll needs to be drastically slowed down so it’s possible to select sensible things.
It’s better with a physical keyboard – Shift-cursor selects stuff in a manner akin to Windows.

Steve:

It gets rather old

Blame avoidance. Other man’s fault, not me mate.
I believe that around 50% of the skill set of management is in finding creative ways for them to make problems they have caused become other people’s fault.

because ROL appear to have spotted the issue and swept it under the carpet.

Makes sense. It’s not the right answer, but an OS change that breaks a load of programs isn’t going to be terribly popular. Note that RISC OS 5 has three modes:
1, crash and burn anything with page zero accesses
2, fake it and log what’s going on
3, bulls**t everybody by just letting it happen (like it has for the past quarter century)

I wonder how many people will get sick of crashes, not understand the huge log file, and will just stick it on the bulls**t mode to get it to “shut up and work”.

Is it just me or is the Filer doing things that should be Wimp?

Such as?

The problem with the RISC OS philosophy is that the core OS services are minimal in places, so a lot of wheel re-inventing has had to happen.

I’d say ripping out certain items would help

Personally, I think there might be milage in having a !User directory, and all the user-specific stuff goes in there – all of Choices, Scrap, ZapUser, etc etc. That way, !Boot can be purely system resources and system setup while User can be all the transient junk and system customisation.
Setup vs customisation? Screen mode is probably setup, it’ll be the same machine attached to the same monitor. Backdrop, on the other hand, is customisation.
I’ll leave it to somebody else to figure out how to implement multiple users…

Such as?

I invite the honourable gentleman to inspect the source file Filer.Keypress and explain why keypress actions substituting for mouse actions (that’s the “m” in Wimp in case you forgot) are in the Filer module.

Personally, I think there might be milage in having a !User directory, and all the user-specific stuff goes in there – all of Choices, Scrap, ZapUser, etc etc.

Choices would be system wide, UserChoices the per-user override of specific items such that deleting the User1-UserChoices sets that user back to using the system defaults, meanwhile User2-UserChoices is untouched and User2 is rather happy about that..

Scrap should be purely temporary files not a dumping ground for files a specific application uses and the app author can’t be bothered to create those in the right place (looking at you NetSurf and friends)

Then the elephant in the room when talking about complexity – the ROxxxHook directories I can see as a logical source of initial files, but not permanent residents, and can anyone give a coherent explanation of why a !System directory has subdirectories relating to previous OS versions back to when the Palaeolithic?
We aren’t loading that stuff across the net with client machines of varying types.

I invite the honourable gentleman to inspect

I wasn’t being facetious (for a change!), I was interested to know what you thought was wrong.

explain why keypress actions substituting for mouse actions

Hmm, all those calls to DecodeMenu_Something… I see. It’s probably because the bulk of the work is done directly within the DecodeMenu_xxx functions, rather than being decoded and passed on to helper functions.
Then, in the future, when keyboard controls were added, it became a choice of “do we simply pass this stuff through to the appropriate menu decoding routines” or “do we rip out all that code, write it propertly as sub-functions, and then everything can call the correct function?”.
The simplest way is to pass it on. It works, and it saves the risk of introducing new bugs.

I do the same thing myself in Manga. Originally, the toolbar and keyboard controls did not exist, it was managed using mouse clicks. So the bulk of the code for handling page changes is in the mouse click function. How does the toolbar and keyboard work for changing pages? It sets up the wimp block as expected for a mouse click, and then calls the mouse click handler.

the ROxxxHook directories I can see as a logical source of initial files, but not permanent residents,

I’m sure somebody will pipe up about the likely miniscule number of people likely to dual-boot RISC OS 5 and RISC OS something earlier on the same machine.

The !Boot supplied with RO5 probably ought to be slimmed down and made “sensible”.

can anyone give a coherent explanation of why a !System directory has subdirectories relating to previous OS versions back to when the Palaeolithic?

Yes.

System$Path : Sys:500.,Sys:370.,Sys:360.,Sys:350.,Sys:310.,SDFS::RISCOSPi.$.!Boot.Resources.!System.

The idea being that the first entry should match the machine (so for RO3.7, it’ll not have “Sys:500”), and it’ll just look through each directory in turn until it finds a compatible module. StreamManager runs on 3.50 or later, so it’s in the 350 directory. And so on.

You’ll note that the Boot stuff and System are offered supporting the Mesolithic era, so I doubt this will change much (although it really ought to).

can anyone give a coherent explanation of why a !System directory has subdirectories relating to previous OS versions back to when the Palaeolithic?

AIUI, it was done that way to simplify matters for the developers, not the users. Hence a module that will work on all versions from 3.5 onwards will go in !System.350.Modules and later RISC OS builds would find it when scanning backwards from their version number. This starts to get messy if later versions of a module will only work with RO5 (so gets put in !System.500.Modules) but all this is doing is wasting disc space as the older version won’t get loaded any more. One obvious advantage of this approach is that it allows for a “universal disc image” that will Just Work with multiple versions of the OS.

You could move away from this approach, but IIRC the scheme was put in place because Acorn realised that upgrading from RO2 to RO3 was such a pain: they wanted something that would continue to function across multiple OS upgrades.

Edit: Ninja’d, curses… :-)

Hence a module that will work on all versions from 3.5 onwards will go in !System.350.Modules and later RISC OS builds would find it when scanning backwards from their version number.

Convenient for one group, not particularly efficient though and as you say “messy”

One obvious advantage of this approach is that it allows for a “universal disc image” that will Just Work with multiple versions of the OS.

Which I would put down as a good reason for a temporary set of directories like the Hook, but unless by some strange means you’re booting several different revision machines from one boot setup via a network it really isn’t required.
Plus, for the last 15-16 years we’ve had a situation where a component (usually a module) can sit in System.370 (or lower) in all its 26-bit glory and have nothing in 500 to load…

How many queries have we had where people broke the boot with a 26 bit item – probably sitting down in 310.

You’ll note that the Boot stuff and System are offered supporting the Mesolithic era, so I doubt this will change much (although it really ought to).

“it ought to”, yep, the point.

Which I would put down as a good reason for a temporary set of directories like the Hook, but unless by some strange means you’re booting several different revision machines from one boot setup via a network it really isn’t required.

Then what do you propose to distribute as a disc image? Or, perhaps, how many different disc images?

Or, perhaps, how many different disc images?

The universal boot, as it is, supported for legacy systems.

And something more appropriate in the one disc image that exists, also replicated for harddisc4 archive.