WPE WebKit

Maybe this is a good solution for a new browser for amigans?

https://webkit.org/wpe/

Everything we may try out in terms of web engines come to question: is it support big-endian. If not, we are out of luck. If yes, it can be possible to follow route Andrea do when first time build first owb versions which was just simple window. For start that can be enough. But problem is official and normal support of big-endianes: if there none, forget about. Morphos devs doing something now with webkit, to make a new browser, but it will be the same out of time because core updating will be pain because of endianes.

@kas1e
On MorphOS they backported some new features (e.g. flexbox support) to the old WebKit engine.

@mufa

As kas1e say the only problem still the big endian support that at some point were "broken" in javascript part of webkit, however the day they fix it (if ever happen) the logical solution would be to update Odyssey rather than rewrite another browser

@samo
I am not about webkit only, but about any possible engine. If there is no endian support then no luck for us. And i didn't mean even endian support for just javascript, but for anything else.

Also forget about idea that even if webkit will have someday big-endian support again (probabaly never), to update latest webkit core to current odysses's gui : morphos devs tryied hard, and have no luck with, so they choice to made a new browser on top of recent webkit (yeah, again new browser). But even then, that mean they need to fix endianes issues AGAIN. What mean, even and if they will made that new browser, they will be forced to be in pain all the time when they will want to update core. It will be always pain in the ass.

Basically, if we found any new core of modern web engine which have big-endian support (that not necesary should be webkit), then create a simple window and render to it will be easy. Then on top of that any new gui can be done in a half of year or in year, which in general nothing and no problems.

The main and big problem, there is no opensource engines with big-endian support which any of us can use. And we have no solution there.

One may say about some "ppc-webkit-port" for macos, but then again, its one-time deal of some already old webkit core. Once you will made some sort of good gui for it, it will be obsolete, and it just never ended loop of problem.

The only solution really is to find out any engine with big-endian support, which will be supported and not dropped for at least few years since start.

Or waiting when netsurf will be matured enough.

@kas1e
I wouldn't bet on Netsurf either. It's a nice lightweight browser, but I can't see it being able to catch up with the modern JavaScript-heavy web. Even mainstream browsers started dropping their own engines in favor of Google's Blink (Opera, Edge). The last major one holding out is Firefox, and even they broke alternative platforms by introducing Rust.

@BSzili
Yeah, i think we all need to realize that probabaly we need forget about some decent supported browser on amigaos4. Just maybe some engine someday by some luck will support big-endian (maybe some phones or some tablet or something will still have big-endianes some day).

I'm not a coder but itc busines anyway. I have been situations like this and at least Netsurf should have a Plan-B status.

It is not a mature enough, but speeding up development is just matter of involved people. If there is a Amiga people involved, there woun't be any problems with endianes ever.

@kas1e
I've been thinking about stopgap solutions, like dedicated programs to use popular internet services. Unfortunately the big companies (Facebook, Google, etc.) are moving the internet towards a more closed direction.
For example Facebook Messenger used to use the XMPP protocol, you could use Jabber clients with it, but now they switched to a proprietary MQTT protocol. There are open source clients (e.g. Pidgin, Adium, bitlbee have facebook plugins), but these are based on reverse engineering, so who knows how long are they going to work.

Other option will be try to make one port of Chrome kit.

May not be relevant at all, but anyway:

https://wiki.raptorcs.com/wiki/Porting/Chromium_(New)
https://wiki.raptorcs.com/wiki/Porting/Chromium/BE

@orgin


IBM has a branch of blink that includes updated BE support.

@terminills
Are you sure? AFAIR IBM's work was for PPC LE, not BE.

Still also SpiderWeb for PowerPC, not based to WebKit but XUL/Mozilla

https://github.com/wicknix/SpiderWeb

@All
Probably some of you may be in interest about browser situations, so there were some post I wrote on os4welt, maybe someone else will find it interesting to read/discuss/etc:


I upload os4 odyssey sources a few times for different developers (do not remember names, one was from .de, another 2 was french ones). I also from time to time upload os4 diffs to some other devs which expresses interest in fixing something, but never-ever anything comes from it.

But yes, I never upload all the stuff to any public repo, which I probably need to do. Today all developers which left on os4 are smart enough to avoid making forks and stuff, so yes, upload to some GitHub will be the right thing to do.

Real issues with any help, is that no one for real wants to help when it comes to something heavy like browser, but only talks. Like "hey hey, where sources? we will update it and fix!", then, once sources in their hands, everyone understand how not easy it all, loses motivation and nothing else. Same as was with Timberwolf, everyone keeps crying "Wtf where sources", brothers put them online on some GitHub, and no one ever-ever does anything with it. They were for long there: https://github.com/tfrieden/timberwolf, now not-avail, but I am sure someone does fork it before somewhere (at least just for some learning purposes maybe). But that all have no value and interest for sure.

Anyway, about os4 odyssey's sources I surely need to upload what I have, even if someone can spot some issue and fix it, it will be good to recompile a new version. While probably no one will contribute, I still should put them on Github, yes.

Now about the current state of things:

Aros version (that one from Deadwood), are the same morphos version from which I do os4 port (so the same possible to port) but contain "a bit" more recent WebKit + aros ifdefs + some fixes (those fixes all in our current version too already).

They have webkit-r187682 (31.07.2015)
Our one have webkit-r155188 (06.09.2013)

So 2 years fresher than our one, but still 5 years old already.

But that repo didn't have aos4 changes as well and it has issues with "endianes" in the javascript core already.

But! Lately, I give a go to check diffs of leopard-WebKit (1-year-old one which working on ppc32 on macs) and then some diffs of some fresh gtk2-WebKit browser (just a few months old one, which also works on ppc32, probably last WebKit which works on) and find out that in terms of fixing "java-core endian issue", there is just simple things to fix. Nothing hardcore.

That means, that with some time and effort I can probably build the latest version from AROS repo with added amigaos4 code/changes, and then fix endian-issue in the java-script core, and it will be a bit fresher WebKit. But it still will be old and dated for today (so those things like youtube and what else those sucking major-popular sites changes today will not work probably as well).

But as those java-core ppc32 endian fixes work for gtk2-WebKit _today_, it also means, that it is still possible to fix java-script issues in the core of today's WebKit and that it still works on ppc32 with some effort. So, theoretically, we can build fresh WebKit, and make it works for ppc32, then on top of it, we can make a new GUI, which is really no problems and just half of the year of time-to-time work of some interested people.

So I tried to download today's WebKit (2gb of shit), and find out there is no anymore pure "Cairo" rendering, they made for each platform some other crap, but at least for Win32 then made "wincairo", so in other words, you can't build core like before, so to just have a Cairo window, and render to it some stuff for tests. You need now to spend some time understanding how current WebKit done, somehow find out the way how to build its core to render to pure Cairo as before, and only after it did, start making GUI on top of it.

Answer on the question "if it is possible to merge the latest WebKit code to our current odyssey (or even to current aros version)": no. Webkit's devs change too much for 5 years. Even on morphos (where few developers work on WebKit), they give up to update odyssey and work on a completely new browser: they the same grab latest WebKit and making ourselves render it over Cairo and start to build GUI on top of it.

In other words, everything looks pretty possible (especially as I know probably how to fix endianness issues with javascript-core) just no one ever from os4 camp was in interest to understand the internals of Webkit and do stuff. Only Andrea before, then Joerg (for him it was easy because Andrea shows that it possible, and there were OWB which deal with all internal shit, as well as in those times there was even SDL and Cairo renderings which you can just enable on compilation stage, not like now), and lately Fab on morphos (i only port what he do with help of others).

Taking it short, that what we have now:

1). re-port AROS version, as it the same morphos version I port before, just with some fixes + 2 years fresher WebKit (but still 5 years old from today). Looks very possible for me to do alone.
or
2). understand how to build fresh WebKit with rendering to pure Cairo and then made a proper GUI on top of it. It will be enough to have navigation buttons and download manager at the beginning, for sure.
or
3). just improve little bits of the GUI, etc in the current version. Not very interesting.


Reporting AROS version for me looks pretty possible, I can do it. I just do not know if we want to spend time on, while its dead end. The right route is port today's WebKit and made a simple GUI on top of it. Very simple GUI for beginning (so just 1-2 days of work), will fit very well for many years :) All what matter is what is rendered to a window, and that you can navigate. For the next few months, it all can be extended and improved, so there I didn't see a problem really. Problem (for me at least on this stage) is to build core to render in the Cairo window because that means working on WebKit's internal code, which is hard (for me, as for non-programmer). If someone else can do it (with my help sure about all other stuff), that can be good. But surely a person who will want to made this should put on hold all other projects, and concentrate on that, and works on it every day, without "ops I have no time, ops I have real-life, ops my cat is dying, ops sorry tired today".

Again, MorphOS developers already build a core of new WebKit of course (they used it in their new mail client) and probably they already working on the new GUI on top of it already and soon will have a new browser.

Maybe (only maybe), we can ask Deadwood to somehow merge to aros version latest WebKit, he is a real programmer (not like me, who only hack-slash-things together), so maybe he can do so. Through if few morphos devs fail with it, then maybe he will have no luck (or even interest) to do so. So if that done, I can port this as well.

That the current state :) But at least _seems_ that java-script endian issues are no issue. At least I know what-where-etc, so if not me, I surely can deal with help from others (even non-amigaos4 devs, like one who does gtk2-WebKit for ppc32, etc).

@kas1e

Quote:

I tried really hard at one time to get OWB compiling but gave up, can’t get it compiling, asked for help, did not get any help, so that was that. Now don’t give a f…k in that project.

Can’t work out dependencies and build system on cygwin, did not compile on OS4. cross compilers and all that crap.

You did get working you soo cool its soo impresive good luck working on on your own.

@kas1e

Considering that since a while Odyssey start to have some navigations problem this is pretty interesting for sure :)

Currently we have tons of good and experimental forks everywhere and frankly speaking it's hard to understand what code is worth and what is purely trash..

Also do you think it could be possible to convince the current developers of the morphos version to share their latest work and then start working ALL togheter into the same github repository ?

For start cleaning things, first thing we may do could be atleast start merging and recollect (in a unique place) all that aros, morphos and os4 code in a single repo :)

@LiveForIt
Quote:

If you can't work out how to use cross compilers, dependencies, build system and "all that crap", then it's not you give a f..k about the project, but project give a f..k about you :)

@samo
Quote:

By navigation, I mean navigation buttons of GUI. Left/right/reload/home and stuff, not the core.

Quote:

Dunno about "tons of good and experimental forks", there just one odyssey code ported to os4 and aros from morphos, and nothing else in working shape anywhere.

Quote:

No, of course not. Why you think those who work on a new browser for morphos, need anyone making their lives harder, who will only replace the date of build, or change some trivial things? I am sure they do not expect any real help from the os4 camp in terms of browser. And they just do not need it. All fine already, they work on what they work and will release it when they did. Without any "All together". They love their babies and do not want them to be touched by anyone else. I can understand it pretty well.

They work fine on what they need without anyone from os4 camp being connected with. They also do not want to be connected with aos4 at all and with all that Hyperion's mambo-Jambo, all issues, legality boredom and with Ben itself because of past. I also remember reading on morphzone that it was told cleary about new browser: it will be only and only for morphos.

So forget it, no one will share the source just only because we want to report their work. Fab was an exception, half of which probably because it's in os4 camp Andrea shows that WebKit can be used on Amiga-like oses, but we already got a bonus before in case of Odyssey's code, so past is past.

If I was on their place, I will share a shit too of course :) Because it's clear that it will only mean they will give their hard work to be used for free on os4, and os4 users will say "only amigaos4 make it possible" or what else dunno. Maybe "mui is bad , reaction is good", etc :) Or there will come some os4 developer who will start "morphos do all bad, it's all wrong, there is proper way to do things". To not have that all, they, of course, will work silently, do stuff and release.

But we discuss it all before with you, what repeat ?:)

Quote:

There is one single repo of old odysseys code done by deadwood already: https://github.com/deadw00d/OdysseyWebBrowser from which can be build and aros, and mos version. I start back in past put os4 changes in that repo too but then just lose interest because of javascript issues. Wawa even tried to build a 68k version from the same repo too. Morphos devs give a fuck about that repo from very begining, so, forget about all those "work together" and stuff :)

Every group of 10 developers and 50 users, will have their browsers and work will be done 3 times differently, yes. The only amiga make it possible :)

@kas1e

Quote:

Do you mean you have problems with the "internal" toolbar buttons to navigate back/forward and so on .. ?

Quote:

Yep i meant all that failed attempts to recreate 68k builds, updating core build etc ...

Quote:

Oh my gosh .. i hoped all that crap was almost forgotten by sane people, still not ? ... do you really think to give all that importance to them ?
No sane people will want continue that absurd camp war in 2020
And today who care about Hyperion, fan boys crap and so on ... and legality of what ?
We are talking about an eventual open source project that could be done by sane and normal developers

Aniway did you ever check the Leopard WebKit project for MacOS X ?
It seems they used a relative modern core based to WebKit-604.5 (more or less a WebKit core from 2018)

@samo
Quote:

No i mean if take a today webkit core from main webkit site, and build it just to render to cairo, then any NEW gui can be done on top of it easyly and that for begining will be enough to just have navigation buttons and download manager.

Quote:

You can't ask morphos devs to open their code. Why they should , even if take aside that they have those reassons ? They do not want to make new browser they works on be opensourced. They do that code, they have their rights for :)

Quote:

When i mean that Webkit change too much, it mean they change too much internals of webkit, and connect it to odysseys's GUI (that one which is amiga gui) are that hard, that better to write new gui.

So in terms of leopard webkit there nothing which is helping there. Leopard webkit together with gtk2-webkit only point me out that fixing of endian issues in java-script core are should be easy. At least they the same for both and for leopard-webkit and for gtk2-webkit, so 2 different platforms with the same fix for ppc32 => should be the same and for us. At least if not, the parts which to check are known.

Realistically issues now is :

1) or to port fresher odyssye from aros repo (so we will have the same odyssey, just webkit core +2 years in compare with our)

2). or to port new-last webkit core and start to made new gui on top of it. Yes, crap about gui, but webkit core is more important. If you have at least few gui buttons to navigate and download manager, and nothing else, but webkit core are of today standards, it mean it can be used. In compare with full-blown gui, and 7 years old core.