New AmiZilla Website

[Trev]

Yes, it's doubtful you'll have a full-featured Firefox running in 2 megs (or less) or RAM.

[Fester]

The new AmiZilla website looks absolutely fabulous.

[Ants]

@ hooligan

Well, AFAIK you can have 512M RAM on WinUAE?

And running on OS3 is really just a neutral base so it can be ported to OS4, MOS and AROS- everyone knows the 3.x API, and can get the docs for it.

-Ants

[hooligan]

Run Firefox on WinUAE while at the same time running Firefox on Windows can be done by pressing alt+tab once? That does not make sense my friend, not at all  :crazy:

[Trev]

@hooligan

WinUAE as a dev environment.

Actually, providing a WinUAE base configuration as a condition for the bounty makes sense for classic development, as it's accessible to almost everyone. It should be possible to put together a pre-built image and build system, a la AmiKit, that prompts for the Workbench 3.1 disk images during installation.

AROS is open but constantly changing, so it would be difficult to define a baseline.

I'm exagerating, but I think the Amiga OS 4 community is something like this: 0.01% - developers with OS4; 0.01% - developers without OS4; 99.98% - prospective users (some of which probably don't even use their PowerPC hardware). Would be developers do not have access to hardware.

MorphOS, at least, is viable in the sense that developers have ready access to hardware.

All that said, Amiga OS 4 should probably be off the table as a target platform until hardware becomes available.

[TheMagicM]

Its for this particular reason why the Amiga failed (3rd paragraph).  A browser for 3.x based systems?  Oh god, please, MOVE ON.   Ditch those old, under specced systems.  

Support whats there.  Sputnik?  Maybe give him some more $ to port that to OS4 (if it isnt already), setup a bounty for THAT project.  Help a browser that exists NOW to advance the real operating systems that are in place, not some ol-timey rinky-dink system.

Remember Amiga games?  What, do you think everyone ran A500's w/512k ram?  Try A500 w/1,2,3 megs, A2000's with 8megs, some with graphics cards.

Good luck, this project will never be completed on a 3.x system.

[codenetfx]

First, kudos to DiscreetFX for AmiZilla site and the Firefox port initiative.

This forum often has too many naysayers - for a forum which is dedicated to Amiga enthusiasts I find that a bit strange sometimes.

Yes, porting a modern browser seems like an impossible task but that's what makes it so appealing . Compromises will have to be made.

I agree that running Firefox on a machine with less than 8MB of ram is not viable. Even if it was possible, memory available after launching the browser would be minimal.

You can think of a single Firefox tab as a 1MB "footprint" in memory. This means that a fully featured firefox could not drive (realistically) more than 1-3 tabs on an Amiga (with 8MB ram). As an experiment, I am writing a plugin (for a PC) which will eliminate the need for multiple tabs (which consume extra memory). The basic idea is that a tab can be represented in two ways: as a memory buffer or as a file (in the file system). Memory buffer won't work on an Amiga because of memory constraints. "File buffer" would work nicely (with a hard drive, of course). This measure alone reduces the need for RAM. I often have a large number of tabs open and think that memory footprint of Firefox is pretty big under heavy use. It does not have to be because I am working with 1-2 tabs at most and the rest are just consuming ram

It is difficult to estimate how big or small the core of ported Firefox would be. Amiga's binaries tend to be very compact but that alone cannot guarantee a small footprint. Once again, I am thinking of a file-system based memory management model which would allow dynamic loading  (and unloading) of components. Yes, this is a bit tricky but not impossible. I have built (and still do!) applications that utilize the same concept on a PC with great success. Each user has its own "usage pattern" for each application and that means that only a few key components are loaded in memory at any given time - which results in very good performance (=low ram consumption).

I am wondering now how many people out there would feel comfortable with looking at Firefox source code and sharing some thoughts here on the forum about memory management and individual components of Firefox.

[paolone]

There are lots of misassumptions in the amizilla webpage, first of all the one about being not considered serious if we haven't our own port of Firefox. Firefox is just another web browser and I personally can think that we would be taken more seriously if we had OpenOffice instead. Someone else can think this about Mono and so on. Firefox has proven to be heavy, difficult, too much tied to features that only "modern" operating system have. On the contrary, some effort on WebKit has already been made and gave us at least some first-stage modern browsers.

Just look at what the only Robert Norris brought to AROS with his "Traveller Preview" or what Andrea Palmatè did with OWB. Webkit is no technically inferior to Firefox and is more viable for an Amiga port, so why insisting on the huge beast? Porting Firefox for the simple pleasure of "having it" is practically meaningless, if the average Amiga machine still working can't run it at all, or with terrible compromises. All those money could be spent in a better way to bring us a modern, portable, light amiga browser that could work on every amiga platform, form 3.1 to 4.1, MorphOS and AROS. More than 5 years without any concrete result should suggest that money are not the issue, here.

AmigaOS 4.1 has Cairo, which is a underlying graphic library well suited for Webkit. Robert Norris ported it to AROS too and, starting from this effort, it shouldn't be so dramatic to port it to MorphOS and other amiga dialects: maybe spending all those money for a more realistic project, as like as continuing and completing Robert's work, might be a better idea.

[codenetfx]

@paolone

This is a great idea (WebKit and Cairo) but it opens an additional "can of worms": AROS, which has limited compatibility with Amiga.

However, if WebKit runs on AROS I am wondering now why it does not run on Amiga already (API compatibility). I have to take a look at that project, provided the source code is available under AROS.

[ajlwalker]

Many people make good points about what is and isn't viable.

However, the problem as I see it is that this bounty has been built up under the premise it will be ported to all these "platforms".

Can you ethically decide to change the target "platforms" without returning the money do the donors?

[codenetfx]

about "changing the platforms"...

The AmiZilla site clearly states that ported browser should "compile and run under Amiga OS 3.1". Since AROS is API-compatible with OS 3.1, browser should run under AROS as well (and probably better, with more memory). In fact, an AROS "box" may prove to be a great development environment.

Where do you see the conflict of interest?

[paolone]

@Trev:

The problem with modern browsers are not in the applications itself, but in the contents they have to handle. Once you succeeded in placing a "beast" like Firefox in 2 megs of RAM (dreams sometimes never come true...), you need to allocate twice of them, minimum, to open a single and light page.

So you need either a) a *huge* amount of RAM (128 megs or above), or b) virtual memory. You can have "a" only on modern hardware running AROS, MorphOS or AmigaOS 4.0, or either under UAE emulation. You can have "b", and for now only theoretically, with AmigaOS 4.1.

Then there is another little problem for AmigaOS: threads. You need them in order to open different kind of contents (music, animations, images, text, javascript, and so) at the same time and, in the end, I haven't either talked about Java.

In the end, there is the graphic API of 3.1 which needs heavy improvements to meet the bare minimum standard to handle a browser properly, and it's the motivation why Robert had to port Cairo to AROS and Andrea had to use SDL to render pages.

@ Codenetfx

AROS compatibility with AmigaOS 3.1 APIs is not perfect but far better than what most people think. In the last months it has improved slightly, and if it's true that there are still many things to do, it's also trye they are far less than they were months ago. Just look at the status page to have a better idea:
http://aros.sourceforge.net/introduction/status/everything.php

There are also some guidelines for coders willing to write their application taking care of compatibility between different Amiga platforms:
http://shinkuro.altervista.org/amiga/guida/english/index.htm

All in all, main compatibility poroblems are between Zune and MUI3.8, but they can be worked around.