general question about riscos.

aaaaa

RISC OS 6 is basically dead today, even if a lot of people would appreciate to get some features in ROS 5 (even if we need to pay for them). ROS 5 is developed today. New things are coming : multiscreen support, better USB and new machines, new memory management, and – I hope – things as FPU/Neon support in C/Basic/etc.

But of course, IMHO, problem is users. We need more users, to get more developers and so a more core developers. From a Linux point of view: 1000 users = 10 developers = 1 core developer. In France, I plan to boost the local market, but also to promote RISC OS as an embedded platform + a great solution for Pi + a cool solution for Basic + a perfect solution for emulation. Hope to prove too that RISC OS could be a good solution for web appliances.

Important point: RISC OS position is quite good compared to the Atari, Amiga and even DOS markets (hey! we don’t rely only on emulators :) ). So we have possibilities to attract some of these users.

aaaaa

Well to get users, it need to have uses :)

To get supporters is enough. I use a bicycle, but it’s just for fun. A lot of people use computers for fun, not to do something useful. We forget that computing is not only business.

What people use computer for is basically well like the video that someone posted, launching firefox and playing video on youtube

So they don’t use a computer, they use Internet. A very few people does computing as a leisure today. Much less than before. Y-generation is a generation of people that always lived with computers, and absolutely don’t care of how they work. Sad. Risky too.

But the way to make this layer, and what it should expose depend in big part of the kind of application and developers that are to be attracted.

We have a rich API with extensions (modules). It’s not ‘Unixish’, nor ‘Windowsish’, but very flexible, because not tied to a specific language. RISC OS IS the framework. A gigantic one.

For ports, I remember my first SDL adaptations. A configure on a PC, to get an example of makefile, then direct compilation under RISC OS. It was pretty simple. With GCCSDK I have the sensation that it’s more powerful, but more complex for a one shot use.

aaaaa

like you could before with BASIC

Why do you think I choose RISC OS? BBC Basic (was one reason).

or that they can take active role in development of things

True.

and having things easy to program and understand is not all that good for business

Not always. There are really a lot of Java projects (or others) that are only stubs to simpler offers (Xojo, VB, etc.). And even more Basic to C or Basic to Java translations. Businessmen ask sometimes for really silly things. Developers are cheating :)

and open source doesn’t change that much with Linux

True too. But the old/modern is not the reason. The reason is monolithic applications VS small tools. With RISC OS, I remember making PNG from some old Sprites: open Sprite in Paint; change background to a colour; drag the result to ChangeFSI to convert it to 1:1 pixels mode; drag the result back to Paint and put transparency; drag to PNG2SPR. Not a single file saved on disc. Just, combination of tools to make a new operation. Of course, some scripting would be fantastic (‘do this’, ‘send result to this’, etc.). Possible evolution for WindOpen?

Anyway, RISC OS GUI is very interesting, because you can make things the CLI way (combine commands together > combine apps together), when monolithic applications try to do anything, even replace the OS. AppleScript was a good way to make almost the same, but is not really used. The RISC OS way seems more natural to people. Drag, drop, redrag, redrop, make data going from one application to one another. It’s very RISC. Big tools are more efficient, but more CISC too, and IMHO, less pleasant.

aaaaa

aaaaa

but not great for filling complex interface , meta information, etc can be a little bit bothering.

Lack of documentation and of high level frameworks… even (especially) for Basic.

that’s also misunderstanding the whole possibility of DOM things.

Yes and no. Modern web browsers are more complex as some whole OS. It’s simply not possible to port them to RISC OS. It would be easier to port an hypervisor and then running Linux. It will solve too the problem of other missing applications (LibreOffice?).

There are ways to partly solve the problem. NetSurf is a very very good document reader. Perhaps that :
1/ some documentation on what is possible with integrated JS could be provided (for webmasters who want to maximise the use of NetSurf features in vertical applications).
2/ some proxy could be coded to intercept call to specific websites and make them compatible with NetSurf (for example YouTube, via a YouTube to Mplayer automatic converter).
3/ more connected applications could replace the web browser when accessing to some popular web services.

For point 3, an automatic launch of third party tools when you go to some specific URL could be cool. A la Flock.

Nota: this 3 solutions are not sexy, but then can be done now.
Point 1 needs knowledge from NetSurf core coders.
Point 2 needs… heu… work.
Point 3 needs some example code of connected apps (for example in Basic).

aaaaa

aaaaa

aaaaa

It’s fine for little play

Yes and no. You can choose to put checks in the framework or application. Complexity is deferred to a higher level. Very RISC like.
Most OS are sets of strict rules. RISC OS is more about open you possibilities. You don’t even have to multitask :)

I have some preliminary engine for an interpreter that can work in both directions: play code in one side, or reverse, self modifying code, etc.
Use? Debugging (SMC to fix a breakpoint in real time, reverse to see what happened, etc.).

Of course, this engine will not to work on a modern OS if I generate assembly and not opcodes. RISC OS is here very useful :)
Security problems (use it to generate complex malware)? Yes. But a lot of new possibilities too.

Or some modules could be made to be matched to html elements, and binded to an interface, and then you can run mplayer inside of the browser.

It’s more a DOM like approach. Very complex, as all the rendering engine of NetSurf must be rewritten (or only parts of it if simple interaction).
Anyway, I think point 1 is the priority. If we can use JS to make better vertical applications, we’ll be able to sell them more easily. More money = more developers for NetSurf.

I could almost use it for the client part of my web projects, but need to provide better interface. Use of internal JS engine could help me. But where is the doc? Where are the examples of working things? Can the engine be compiled as a standalone application for server use under RISC OS?

A big part of Firefox development was made or paid by companies that use Gecko for vertical applications (for example Peugeot in France). The same for Linux. Users will not give you money. Latest example was JPEG update for RISC OS. Paid mainly by a developer who need it for its software.

aaaaa

aaaaa

aaaaa

Modern web browsers are more complex as some whole OS. It’s simply not possible to port them to RISC OS.

It is certainly a large amount of work. It is certainly very complex. But it is certainly not impossible. WebKit was ported to many different OSes including MorphOS and AmigaOS.

If you insist to use the Norcroft compiler for such a project however, it might well be near impossible.

aaaaa

aaaaa

Check your hardware. Neither IE nor FF nor Chrome have ever crashed my Windows machines. Broken network drivers on the other hand…

I used FF for a long time on an Atom-“powered” machine. Slower CPU and slower RAM than the PandaBoard. It worked OK for general browsing, only Flash was a bit slow. And it was 100% stable.

There are many reasons why WebKit or FF won’t be ported to RISC OS. But it is not memory leaks/consumption/performance or OS limitations.

The longer I discuss these things, the more I am motivated to do a quick hacky minimal WebKit port. Just to prove the point.

I am motivated to do a quick hacky minimal WebKit port.

GO FOR IT!!!!!

There are many reasons why WebKit or FF won’t be ported to RISC OS. But it is not memory leaks/consumption/performance or OS limitations.

I have to agree. Firefox is working fast and stable with Angström on my Pandora. The Angström Netsurf version is slower than RISC OS Netsurf on the same Pandora. Ok, I should not compare this.
The Pandora Firefox is comparable with a PortableApp in Win. All inside. The source is aviable, I think.

And there is port and port as well.

Exact. Probably a port without canvas, without webgl, without websockets, without JIT, without video, without sound, without etc. Summary: without HTML5. Was the same on Haiku. WebKit, but so limited that they stop the port. And they have QT.

The longer I discuss these things, the more I am motivated to do a quick hacky minimal WebKit port. Just to prove the point.

Why not :)

Hi, guys, was just reading your thread and had a random idea, forgive the coarseness of it – What about porting Lynx/Links and incorporating that minimal webkit renderer on it? I mean, I personally would rather develop straight into RISCOS using RISCOS mentality – do what needs to be done and do it well.

Links was quite snappy from what I remember, could that code not be used to support modern HTTP protocols, SSL, etc whist rendering CSS2/3 and HTML4/5 with the WebKit engine?

I like the idea of using HTML/CSS/JS for UI descriptions, validation, etc. But how would we access the interface from the back-end ie in BASIC? Maybe with a DOM-like SWI? Hmmm…

Maybe with multicore CPUs, we could render the WIMP on one core and execute the main code of programs on the other cores?

I’d probably need to see a snapshot of RISC OS as it is currently, but there should be a holistic method to secure and maintain correct (ie the usual lightening fast we have come to expect from RISC OS) performance without just adding more code… I also like the fact that the TCP/IP stack is separate from the system.

Please let me know what your thoughts are :)