Why is it that Web browsers seem to be one of the things that are just not ported whereas they are one of the most needed items of software? Are they technically very difficult to port?

i'm perfectly happy with !Oregano. Porting is *very* difficult, usually, most people give up soon (I vaguely rmember something called Rozilla???)

Why is it that Web browsers seem to be one of the things that are just not ported whereas they are one of the most needed items of software? Are they technically very difficult to port?

The vast majority of them rely on features such as pre-emptive multitasking and multithreading to help with network management. Porting to RISC OS would probably involve significant reworking of the browser's structure, either that or writing a threading implementation which, I believe, was the route that Rozilla was going to try.

In terms of difficulty they probably represent some of the more difficult things to port. Some idea of the effort that would probably be involved must be seen in Oregano 2 which supposedly builds some kind of abstraction layer to get around this. How this works where the features simply aren't available I've no idea. Perhaps this is one of the reasons that Oregano2 is taking so long.

Getting a browser working under a different windowing system is probably one of the smaller concerns. All window systems tend to have similar features, although it will be difficult if the code to manage the UI is tightly integrated to the main browser code.

[Edited by johnstlr at 16:15, 12/4/2002]

Actually I think the author wants to do a complete rewrite to structure the code in such a way as to make dealing with this in future easier than it is now.

(Ooh heck, what have I gotten myself into here?!)

I think if we got enough people together with enough combined skills, we might actually be able to do this. I can do plenty of BASIC, and bits of ARM assembler and C/C++ as well as Python (probably not suitable for a web browser!)

http://grail.sourceforge.net/

It seems that BASIC has given interpreted languages a bad reputation.

Hardly surprising given that most flavours of BASIC were / are terrible

Grail uses the select module to achieve asynchronous networking.

I guess the select module maps down onto a standard socket select call. It's not truly asynchronous, rather it allows a single thread to monitor a set of sockets for activity rather than just one.

In any case calling it will block RISC OS. In fact I'm pretty convinced that the only realistic way to achieve multitasking networking on RISC OS is the method I outlined in another thread which would probably mean alterations to the network management of any application. Sad to think that, in this respect, RISC OS is on a level with Windows 3.1 eh? Actually what's worse is the fact that the development tools don't even support this way of working as standard - you have to write the module yourself (or find a 3rd party one, I know of at least one).

I wasn't seriously suggesting that someone port Grail, just pointing out that one can write a reasonable web browser in Python.

Fair comment - sorry I took it a bit too literally.

Thanks for the links BTW.

Thanks for the corrections.

No problem.

Grail uses the select module to achieve asynchronous networking.

I guess the select module maps down onto a standard socket select call. It's not truly asynchronous, rather it allows a single thread to monitor a set of sockets for activity rather than just one.

In the absence of multithreading, it's the best you can expect, I'm afraid.

In any case calling it will block RISC OS. In fact I'm pretty convinced that the only realistic way to achieve multitasking networking on RISC OS is the method I outlined in another thread which would probably mean alterations to the network management of any application. Sad to think that, in this respect, RISC OS is on a level with Windows 3.1 eh? Actually what's worse is the fact that the development tools don't even support this way of working as standard - you have to write the module yourself (or find a 3rd party one, I know of at least one).

SocketW@tch[1], I presume.

I wasn't seriously suggesting that someone port Grail, just pointing out that one can write a reasonable web browser in Python.

Fair comment - sorry I took it a bit too literally.

No problem. I wasn't aiming the comment at you in particular.

[1] The method used to catch offensive language on this forum just draws attention to places where they are inadvertently contained within other words.

In the absence of multithreading, it's the best you can expect, I'm afraid.

Not entirely. On unix, AIUI, it has always been possible to get an asynchronous callback from the OS when there was activity on the socket. The same calls are available in RISC OS but can only be exploited (in C at least) if you write a module. Of course this still leaves the problem of how you get the network event to the application.

SocketW@tch[1], I presume.

That's the one.

This is what socket**tch does. The point I am trying to make is that this is not how most network applications on other platforms are structured and thus porting them requires, probably significant, changes to the application's internal architecture.

Is that explicit enough? Sheesh.

Ah, I understand now.

SWiRO, ermm, no dunno

of course generating an acronym would probably be trivial.

or muff diver, heh.

but crowbar sounds good to me, crows being cool and black n'all.

How about naming it after what you do with it? Draw, Paint, Edit, uh, Surf? Some thoughtless swine has already used 'Browse'.

Hmm. 'Net' and 'Internet' have already been used up and I can't think of anything decent using Web, WWW, or Network. Interscape? Net explorer? Anyone?

Vector was named after the basic unit of vector drawing. How about !Link? Erm. !P? !Marquee?

Or Web Anchor. WAnchor for short. Urr, then again...

How about a web browser as a window or guide to the internet? Noone suggest Internet Navigator! Ancilla? What's a posh word for window?

[Edited by ninj at 20:48, 18/4/2002]

One major concern I have is that the rendering and graphical features of RISC OS browsers is behind Opera, Netscape and Internet Explorer. I would be very interested in developing this side of the browser.

There has also been much discussion on this thread about networking (something in which I do not have much experience). Several individuals have suggested it might be hard to get that side of the browser working. One of my ideas (this may be completely useless - I don't know) is that we could use existing modules to retrieve the pages from the internet - so allow us to concentrate on the HTML p@rser and rendering engine.

People-wise, I think we have:
Me: Coding
MarcoF: Testing
Monkeyson: Coding
Andrew Poole: GUI Templates
Moss: Testing, Docs

So, to conclude, yes, I am serious, but think we ought to concentrate one thing at a time - a web browser is a big project.

[Edited by Hertzsprung at 09:20, 19/4/2002]