The Elephant in the Browser

The alternative to Steve’s response is also to note that clinging to the past for far longer than is reasonable will affect us moving forward. I am aware that there are probably still people using A-series hardware to access the Internet. Heck, there’s probably somebody still using ArcWeb if you look hard enough.

The problem, at the other lot also found, is that you can make whatever improvements you like but there’s not much point if the users don’t follow.

My gut feeling, based on talking to people both at shows and at user groups, is that this forum (or even the newsgroups) isn’t really representative of those using RISC OS in the wider sense. I think we’re all in agreement here that we should support new hardware and modernise the OS; the people who you’ve got to convince are those who attend the shows and earnestly explain that they’re still using RISC OS 4.02 on a RiscPC, or RISC OS 6 on Virtual Acorn, and have found that some new piece of software or other doesn’t work on it.

The sad reality is that, away from those of us who like working on the OS and its software, people aren’t still using RISC OS because it’s cutting edge or modern¹…

¹ They’re using it because it’s familiar, and because someone told them in the early 90s that Windows was evil.

Steve

Whilst some of that is true I’m not convinced RISC OS alone suffers from people sticking with familiar as witnessed by those who refuse to upgrade from XP to Windows 7 and 8.×.

Also I’m sure, speaking to those users I know, that there is a desire from users and not just developers to see RISC OS improve and that some of this is seen with the take up of Raspberry Pi’s.

Where ever possible users feedback should be taken in to account and most certainly when major changes are proposed as that way hopefully more people come on the journey.

Whilst some of that is true I’m not convinced RISC OS alone suffers from people sticking with familiar as witnessed by those who refuse to upgrade from XP to Windows 7 and 8.×.

Not upgrading from XP is not something I’ve encountered away from the RISC OS world: most people are happy to upgrade to Win 7 because it’s more stable and secure than XP (the latter being the thing usually cited as a reason for not upgrading around these parts).

Win 8’s less popular, but then that’s just following the Windows tradition of alternate releases being a bit rubbish. The current signs are that Win 10 should be better, and people I know are sticking with Win 7 and waiting for 10 to become available on that basis.

Not upgrading from XP is not something I’ve encountered away from the RISC OS world: most people are happy to upgrade to Win 7 because it’s more stable and secure than XP (the latter being the thing usually cited as a reason for not upgrading around these parts).

In some ways the XP and RO 4.x issues are rather alike¹. However, in the RO world the newer faster boards required for the new release don’t cost a fortune. There’s way less of an excuse for not buying a Pi2 (and housing it in an RPC case² if that keeps you happy)

Win 8’s less popular, but then that’s just following the Windows tradition of alternate releases being a bit rubbish.

As someone who has to work with other support staff who do deal with server and desktop installs I’d take issue with the word “bit” in that. Substitute “lot” and you get the idea.

¹ Many businesses and government agencies (not just the NHS and not just the UK) have critical XP reliant applications that simply haven’t been updated. In some cases the updates exist, but another critical application has no update and often the “updated” item is actually a different beast altogether with no backward compatibility thus leaving the user needing two machines.

² Use a spare drive bay and pass a handy 5V line to the board. I wish to state this is my idea so trying to sell me one later won’t wash :)

Hmmm? I know of over 4000 XP machines with updates and patches issued at regular intervals.

Yeah, and the taxpayer is stumping up how much to keep the NHS’s XP machines running?

I am using XP on an old netbook, and I have just installed it on an older (but better) computer. The decision to use XP is part the age of the hardware and part because the best driver set for the hardware is for XP/Vista, plus some of the devices I use frequently (TV capture box) don’t work post XP.
I’m not saying XP is best, it’s only that it is what works best for my needs but I wouldn’t expect the PC market place to continue supporting my machine for all eternity. ;-)

and the taxpayer is stumping up how much to keep the NHS’s XP machines running?

Like I said, not just the NHS and not just UK. Legacy apps and all that.
Let’s not insist that RO has to hang on to all that baggage – not least because the often quoted “It isn’t 32 bit” is:
1. Often barely true – just the other day Dave Ruck pointed out that an item labelled as not 32 bit merely needed a new header.
2. Where true quite a lot of the non-32 bit items are actually relatively simple to modify to 32 bit compatibility. This last comment based on my own hacks. Needless to say, where the source isn’t open a hack in my house is no use in yours.

Many businesses and government agencies (not just the NHS and not just the UK) have critical XP reliant applications that simply haven’t been updated.

Indeed: I’ve got a few boxes at work that are still running XP Embedded. I was referring to individuals not upgrading, since I can’t believe that the sorts of problems that affect the NHS or companies using customised PCs to control manufacturing equipment commonly affect the average home user writing a parish magazine or watching amusing cat videos on YouTube.

1. Often barely true – just the other day Dave Ruck pointed out that an item labelled as not 32 bit merely needed a new header.

That’s only the case in some very specific circumstances.

affect the average home user

Indeed. For them, the off switch was probably the fact that Windows 8 didn’t look like Windows. Even now, playing with 8.1 in shops, it seems harder than it should to switch the thing to look like it ought to look instead of a bastardised tablet style interface where it seems depressingly incapable of multitasking stuff side by side.

That’s only the case in some very specific circumstances.

Indeed, we only need to look to Impression to see that making something “26 bit” become “32 bit” may be considerably more of a challenge than it looks. And that’s with source code.

Thankfully Zap has been converted (though it is quite buggy) as has OvationPro and DigitalCD so, well, that’s most of my necessities sorted. ;-) There’s some stuff I’ve had to leave behind, but that’s how things are. Sometimes you just gotta move forward. With the Pi being as inexpensive as it is, and RISC OS 5 freely available, it would probably pay for itself in the electricity saved in not running a big desktop machine. Really, my RiscPC has a ~100W PSU plus a secondary 240W PSU for the additional harddiscs. The Pi? Runs off a decent quality tablet charger and is totally silent in use. What can I say? ;-)

That’s only the case in some very specific circumstances.

Er, yes. The circumstances where someone didn’t use 26bit only code constructs when creating a module or similar before the information about the flags word was common knowledge.
I was quoting an instance where the situation hit an extreme. As I also said others have very little 26bit only code and literally running the code through !SID with the macros setting selected, adding the extra header lines and then recompiling for 32bit is all that is required.

Best to take each case on its own merits and avoid sweeping generalisations.

I never said complicated applications were simple to convert. Rick quotes Impression, which if I recall correctly didn’t have fully documented source and may not have been full source. It used various tricks that don’t work for 32 bit mode and the new maintainers are apparently adding new features. Not simple at all.
Rick has in the past mentioned the HENSA archive and the “lost” items in there. I wonder if anyone has actually checked how much is still useful and how much contains few if any code problems.

There’s a certain amount of “Chicken Licken syndrome” about some of the statements.

Is the Epiphany Browser relevant? http://www.raspberrypi.org/web-browser-released/

As far as support for Risc PCs goes, it would be nice for programs to be usable on a strongarm, (and work if you don’t mind waiting on an A7K) because that means they should work well on all new hardware.

And as for political reasons for sticking to RO 4.x, isn’t it still the case that there are some features that RO 5.x is still behind in?

I was referring to individuals not upgrading, since I can’t believe that the sorts of problems that affect the NHS or companies using customised PCs to control manufacturing equipment commonly affect the average home user writing a parish magazine or watching amusing cat videos on YouTube.

I know plenty of individuals who haven’t upgraded from XP.

The main reason is that, in their eyes, while the computer still works and runs their software (mostly just a web browser to access Facebook to get their fix of cat videos and photographs of people’s lunch) there is no need.

In some cases, if pressed they’d probably offer another reason along the lines of “can’t afford to” (Not that a new computer has to be expensive, but if they say it’s too expensive, then for them that’s what it is.)

And as for political reasons for sticking to RO 4.x, isn’t it still the case that there are some features that RO 5.x is still behind in?

As I said earlier: “Silly as it might sound a few tweaks of the GUI (key shortcuts) seem to be a key missing component”

I know plenty of individuals who haven’t upgraded from XP.

Come to a decision: Since I have a machine suitable for Win7¹ the win2k machine can take its pension. Upgraded from XP?

¹ Just need to make sure it can’t run farmville or whatever that stuff is and I won’t end up with complaints about the machine being “slow and crashing and…”

As I said earlier: “Silly as it might sound a few tweaks of the GUI (key shortcuts) seem to be a key missing component”

I was thinking more of the image render stuff. (Although I don’t think it was ever really used properly by many apps)

Is the Epiphany Browser relevant?

Only if a relatively recent GTK+ has been ported to RISC OS, and the point of Epiphany is that it’s a GNOMEy WebKit-based browser. And, it appears that it uses JavaScriptCore (or, as Apple calls it, Nitro), so it’s not even doing anything special with its WebKit, AFAIK.

I was thinking more of the image render stuff. (Although I don’t think it was ever really used properly by many apps)

I had rather dismissed that for the very reason you gave. I think it sits slightly higher than the Appletalk¹ protocol support though.

¹ Horrible network protocol that ranks alongside IPX and NetBEUI, down in the pits

It felt odd when ROL added AppleTalk support.
A quick check – http://en.wikipedia.org/wiki/AppleTalk – and AppleTalk was unsupported as of 2009

The Image File Render stuff never took off because, like all Select enhancements, it was limited to Select and latterly Adjust/RISC OS 6 :(

AppleTalk was unsupported as of 2009

I wish printer manufacturers (like Ricoh) would leave it out of their current kit. Having it in and on by default is a pain, that and the equally naff Bonjour.

The Image File Render stuff never took off because, like all Select enhancements, it was limited to Select and latterly Adjust/RISC OS 6.

Lets open the wish pot again and hope A.T. opens the source.¹ It seems like there are just a few basic items of Select that people say they can’t do without. But there’s a basic feature of RO5 they can’t do without long term.

¹ More likely I’ll win the lottery (and I don’t play)

Accept that the RiscPC is ancient, and is not an acceptable target for modern software.

Admittedly I’ve never used RISC OS on anything other than a Pi, but I’d agree that targetting ancient hardware can only hold the platform back. I was amazed to find that the 2015-edition Style Guide suggests supporting as far back as RISC OS 3.5 if you can, and also advises that you “consider” supporting 3.1. Both of those systems are older than Windows 95, and nobody makes software for that anymore!

It’s even more insane when you consider that anything capable of running 3.5 is capable of running 5.21 (although some hardware may not be supported properly – 26-bit podule ROMs).

Here’s the minimums, I’d say, for what to support:

If you want to support modern stuff, whatever the current ROOL version is. Everything that the 5.xx branch has ever supported is currently supported, and it’s even backported to IOMD (RiscPC and A7000) machines. IIRC you’re best off having a 4.02 or 4.39 ROM to softload from, but still…
If you want to be compatible with 26-bit stuff, 4.02 minimum (but don’t use long filenames). AFAIK, even today, it’s not hard to get 4.02 ROMs, and you can softload to 4.02 from 3.6 or 3.7.
If you want to be REALLY compatible with old 26-bit stuff, 3.70 minimum. Really, does anyone even run 3.5 or 3.6?
If you insist on being compatible with the ancient pre-IOMD stuff, 3.10 minimum.

Mind you, if what you’re doing doesn’t actually benefit from newer stuff, no reason to NOT support the older versions, but…

Still completely true for any software that wants to be taken up by non-technical users. Software can’t rely on third-party hacks like Wimp2.

Taskwindows (available), Light threads (see Firefox).

Aside from the SCL, which doesn’t count in the context cof the NetSurf info, are any ready for rolling out to non-developers?

GCC4 supports them, by default, as ELF (and yes, I don’t like that).

Yes: it it doesn’t, that’s a lot of the current RISC OS users who are suddenly unable to use the browser.

I did not say ‘available’ but ‘supported’. Not the same :)

The alternative to Steve’s response is also to note that clinging to the past for far longer than is reasonable will affect us moving forward.

And so the question: should we still support officially ROS4? With so many new technologies under ROS5, I would probably make two branches. One for each OS.

Anyway, There is another approach: API. Thanks to the mobile world, web applications are almost all available as API, so you can make applications for them. You can basically use an Android phone online all the day without a web browser. That could be a solution for RISC OS. ‘Just’ need a better and simpler way to create software.

With Xojo, I can make a GUI application connected to a web service in minutes, and finish it in less than 3 hours. The same for RISC OS would change a lot of thing.

The point is that web sites that do not work under RISC OS are always the same: Youtube, some news services, banks, social networks. Not a lot of applications to write. And that will be simpler – IMHO – than to try to port WebKit to RISC OS (face it: it’s technically impossible). Use the web browser for content, and native applications for web services. The web browser could even propose you to use some native applications when you’re going to a specific place.

Not a lot of applications to write. And that will be simpler

Apps are usually feature poor compared to the website. Good luck getting an app API from a bank, and doesn’t youtube change how it works rather often?

Getting an API from the banks would be the simple part, getting it authorised would be the interesting/impossible part.
(And no, it wouldn’t really be simple getting the API)
I’ve been involved with multiple banks integrating with their API’s it’s at least 6-12 months full time work.
So it’s definitely easier and quicker to write a single application than a web browser.
But if you consider the number of web services out there that people use the best API is html.

However getting applications for twitter, facebook, youtube and a good rss reader would make RISC OS more usable and be realistic.

The only reason you can use an Android phone like that is because there’s a business model for businesses to develop single-service apps for Android.

There is not such a business model for single-service apps for RISC OS. Why? Because there are four of you

It’s also worth noting that the single-service apps on many platforms simply embed a web browser in a way that looks more native, especially on more niche platforms like Windows Phone. For that matter, the YouTube app on Windows Phone is just the mobile YouTube site without an address bar. And, this is Windows Phone, which has Microsoft backing, and Microsoft is big on developing apps for businesses for free for the sake of platform adoption. We don’t have Microsoft resources, we can’t do that.

Then again, banking on RISC OS is a really, really bad idea.