Any interest in a crowdfunding for a WebKit based browser for OS4?

[NinjaCyborg]

I work with a team of professional webkit porting experts. Wondering if there's any interest in raising a fund to pay for them to port WebKit to OS4 (OS3 is probably a stretch even on a vampire, partly because of RAM requirements and partly because Cairo lib and some others are not available)

[NinjaCyborg]

I agree ongoing updates is a problem. However, unlike in 2013 or 2010 or previous, web technologies have stabilised and matured - HTML5, CSS3, ECMAScript 6, websockets, webgl and so on. So, a good port now should last a long time even if it didn't get regular support. Also the guys I work with maintain official webkit ports for a major embedded platform so hopefully that work could sustain ongoing updates.

[NinjaCyborg]

Hi all I wanted to accumulate a few replies before replying.

• Budget would likely need to be high 5 figures, realistically
• Skillset - this team is very, very good and there is Amiga (and even 68k assembler) knowledge, albeit not up to date
• JIT - their other versions don't use JIT, there isn't a good Big endian JIT implementation to lean on
• Yes it would remain open source, unlikely that it would be accepted back into upstream though
• UI - here's a thought, what about using the existing AWeb UI but replacing it's rendering engine and JS library?
• A Patreon model would work better than a bounty or kickstarter

[NinjaCyborg]

Obviously, if we're going to do it we'd publish a 'prospectus' of why the team is credible and do some initial work to prove it.

[NinjaCyborg]

I see a few possibilities here:

a) NinjaCyborg is just trolling for lulz

b) he hasn't done his homework in terms of what is available already, what would be needed and what wanted

c) he has done his homework and came to the conclusion that the crowd around here is gullible enough and rife for another picking

Go **** yourself

[NinjaCyborg]

1) The WebKit team I work with maintains the PlayStation webkit port for all their consoles
2) I ported Python to Amiga in the distant past and to Symbian OS
3) I worked many years at Symbian and know a lot about operating systems development, managing big software projects and porting browsers like Opera and WebKit to obscure systems. I've worked for some of the biggest names and run some of the world's most popular websites.
4) I have a strong fondness for the Amiga despite having been off the scene for over 15 years

However, if this is the reaction from Amiga community to having nice things, you can go **** yourselves

[NinjaCyborg]

@wawrzon Nobody is asking for any money now, I just answered realistically to a question about what the budget would need to be. Perhaps you all don't understand how crowdfunding works, but no one person is going to be asked for five figures. If there were 5000 active amigaos4 users, who all paid $10, that would go a long way to getting professionals involved, not the amateurs you've had in recent years. And these are professionals - they write the YouTube, Hulu, iPlayer and countless other web and VOD apps you have on your PS3, 4 and Vita.

As for OWB, it's wildly out of date. The point is to give you all the ability to run html5, google docs, office365, gmail and even write new apps using an amiga version of Electron.

But hey, if you guys don't want it, that's fine, that's the message I'm getting from the market research I'm doing here. You'd all rather just snipe at each other.

[NinjaCyborg]

No one is asking for any money before proof of capability. What is wrong with you people?

[NinjaCyborg]

@Hans_ not ruling anything in or out, was trying to have a sensible discussion until a couple of arseholes hijacked the thread. - Thanks for bringing it back on topic by the way - OWB is so out of date it's probably not just a case of updating the engine. But we'd investigate it for sure. Remember - we have an existing up to date WebKit/JSKit engine port that runs on embedded PowerPC systems with few dependencies on OS provided libs, that we can potentially use. As for suggesting the old AWeb UI, it's as good as any, it was just a thought. Since having a proper Amiga UI is preferable to say, the way the NetSurf-SDL version handles the UI.

[NinjaCyborg]

@ferrellsl I have been off the scene for over 15 years it's true, though I followed the news enough to know about TimberWolf, NetSurf, OWB and OS4 drama. This account is new, because I have long since lost my old ones and associated email addresses. I have been cynical of the Frieden brothers since ~2003, the reason there has not been a 68K update of Python in all that time is because the Frieden brothers refused to share the changes they made to Python for OS4 back to me to include in the 68k build, despite building upon my work. Since I had plenty of non Amiga things to do at that time and since they put me off from contributing further, I didn't.

Now I have time to spare and the release of 3.1.4 reignited my passion for Amiga, and the contacts I have from years working in the operating systems world put me in a position to be of use. But, whether we do something or not really depends on whether the community deserves it, or it just going to keep flinging turds at each other.

[NinjaCyborg]

@cgutjahr LOL get over yourself

[NinjaCyborg]

@Spektro The datatypes system is not really sophisticated enough to support a proper web datatype, for example inline images was part of the AmigaGuide spec, but never implemented, as it can only cope with data file at a time, no child data files. Datatypes is also missing data stream and progressive rendering support that makes it unsuitable for us with a browser compared to simply embedding the raw codec libraries. Admittedly it's not very Amiga like to have duplicate code, but it's what every other system does and we certainly can't rely on the OS getting 21st century features anymore. BUT what I was thinking was as follows:
- an architecture that makes both the web engine and the JS engine available as separate entities (imagine being able to use Javascript for rexx scripts)
- a web browser that obviously embeds the two
- a light quickviewer suitable for previewing files, that again just embeds the web engine. This would meet your 'datatype' need for use with workbench, DOpus etc.

This is what webkit is architected for, like how it works on the Mac between Safari, the system WebView and the Finder QuickView. It would make it possible to have Electron which in turn is what many popular apps like Slack and Skype are now built with.

This even leaves it open to other browser UIs if you don't like the default one.

As for 68K support, I already mentioned while it would be great to be able to do, the combination of missing dependencies like cairo, lack of recent version of GCC, and the need for 200MB+ RAM would seem to rule it out.

[NinjaCyborg]

@gregthecanuck
I completely agree. The purpose of my original post here is to gauge whether it's worth pursuing before spending time and money on the initial risk mitigation and proof of concept work. Out in the real world, I am a published author on the topics of lean/agile software and project risk management. And I was once commissioned by Mozilla to investigate porting Firefox to Symbian OS. The Firefox code was a dumpster fire at the time (it was heavily refactored in recent years). So I'd like to think I know what I'm doing.

[NinjaCyborg]

@Spektro totally agree a way to embed the engine would be a priority and the browser itself should be architected in that way. Datatypes are probably not the right way to do it based on my deep knowledge of OS3 and shallow knowledge of OS4 but BOOPSI might be simply because BOOPSI is pretty bare bones as it is. I can imagine a child type of the reaction window.class that implements a webview window as one possible approach.

[NinjaCyborg]

@wawrzon I'm already being accused of making promises I'll never keep, despite not yet making any promises, so I definitely won't promise a 68k version! I'd like to think a vampire build might one day be a possibility but right now I'm just spitballing ideas.

I have had a play with OWB on AROS and it renders well but certainly uses all the ram it can get. On OS4 it's just too unstable IMHO. Again, we already have an embedded build for powerpc, and frankly working with code we already know and can build upon is likely to be more productive than importing code from another side project. WebKit also needs a recent GCC 4.7+, which I don't think exists on 68k but maybe exists as a cross compiler?