Amiga.org
Amiga News and Community Announcements => Amiga News and Community Announcements => Amiga Programming and Development => Topic started by: deadwood on January 06, 2016, 06:49:40 PM
-
I would like to invite interested developers from all platforms to join the Odyssey Web Browser development.
The repository is public, available to everyone and located on GitHub:
https://github.com/deadwood-pl/OdysseyWebBrowser (https://github.com/deadwood-pl/OdysseyWebBrowser)
The repository contains the latest version of Odyssey (1.25) which:
- is a continuation of Fab's 1.24 code base
- is rebased directly on WebKit engine (no longer uses Orygin proxy layer that 1.24 used) for easier integration of new engine versions
- uses WebKit revision 187682 (end of July version)
- is rebuild to support multi-platform development in build configuration and source code
If you are interested in developing a feature in Odyssey, no matter how small, or fixing a bug, no matter how trivial, contact me via email at deadwood@wp.pl. Don't worry - everybody had to start somewhere! I will help you understand the code base, build process and development approach.
Please spread the news!
-
Deadwood is great developer. And now AROS has the most updated and current Odyssey in the world of Amiga.
In my opinion all NG systems must have a new OWB, modern OS without good browser is nothing. Can we do it beyond divisions between camps? We need it!
-
There is with OWB. It's for all camps now. The problem is that a core component doesn't work with big endian anymore, and the webkit team don't care to fix it. This means that webkit doesn't work with PPC anymore. That is why we havn't seen an update in 2 years except for little endian AROS. It's not infighting, it's a technical problem. And the guys are trying to find a solution for it.
-
deadwood proposed a bounty for big endian fix few weeks ago but scrapped the idea due to the apparent lack of interest. here you go, if none speaks up.
-
It's not lack of interest, it's lack of developers who can fix this problem.
-
so deadwood is not good enough for you?
-
I didn't get the impression he would fix it. Only that a bounty was proposed. If deadwood can fix it I bet a lot of people would donate. Me included.
-
deadwood has proposed a number of bounties, among others exclusively for aros68k. what would be his interest to propose those, if he was not intending to eventually pick them up? i mean, im not sure, but he likely is like most aros developers, mainly an x86/x64 user. its unlikely that he needs that endian fix himself...
what you say is the basic problem of the bounties, users will not donate if a trustworthy developer doesnt volunteer in advance. developers dont know how high the interest of users is. both is understandable. but someone tries to estimate interest, good thing would be to speak up, if you want it to take off.
some trustworthy site like people2power should have polls to estimate such bounty interest in advance to actually tarting them.
-
Maybe a solution is to create a Odyssey Team? I'm not developer, but I see it like core team could prepare release of Odyssey in every new year. Something like with IBrowse - new year = new version. Than community can support team with bounty every year or buy for example new version for 50€ every year. It could make sense, because summary of NG users is not so small, could be enough € to sponsor our developers.
-
The solution is to find someone who is capable and willing to fix JavaScriptCore to work on big endian processors again, and pay him to do it if necessary. This could be done on Linux or Mac OS X, so he won't even need OS4 or MorphOS to do the job.
-
Stupid question, but how we can find someone? Inside our community or better outside?
Or maybe we can find other alternative for WebKit.
-
It might be worth looking outside the community, like they did for the H.264 altivec optimizations.
-
Who should we contact? I now, again stupid question...
-
The solution is to find someone who is capable and willing to fix JavaScriptCore to work on big endian processors again, and pay him to do it if necessary. This could be done on Linux or Mac OS X, so he won't even need OS4 or MorphOS to do the job.
Why not just fill a bug report to the webkit team, they should fix what they broke right?
Kamelito
-
Why not just fill a bug report to the webkit team, they should fix what they broke right?
Kamelito
certainly, the big endian users have the highest priority with them. thats why they have practically abandoned to support this byte order. similarly why not complain with mesa/gallium team.
-
Why not just fill a bug report to the webkit team, they should fix what they broke right?
Kamelito
Fab did and was told iirc "we don't even own a ppc machine so you'll need to do it yourself"
-
What a damn shame. PPC had an amazing browser on MOS and its dead in the water now because of silly things like this. What about a 68k solution? Same issue?
-
In this situation or we must do it in our community or maybe jump to other base?
I wrote a small test of Qupzilla, browser for QT: http://www.amigapodcast.com/2015/11/qupzilla-for-amigaos.html (http://www.amigapodcast.com/2015/11/qupzilla-for-amigaos.html). Maybe this is a solution?
Or jump into Firefox, but our sources I think are completely outdated.
At the end of all we must have solution of this problems for NG systems.
-
Odyssey is not the only option per se, but the question remains the same: who is going to do the work?
-
Would it be stubid move to start supporting Netsurf? Not just one port, I mean whole Project? Netsurf is developed for nieche platforms and I could imagine it would be a good thing many ways? At least some problems what devs get with webkit, becuase it is major player would be history. I would bet that there will not be endian issue with it ever. There is a team and this way it wouldn't be too much for just a one people.
There is a HTML5 parser on the works http://www.netsurf-browser.org/projects/hubbub/
Java is partially working already etc.
-
Netsurf is great for platforms with limited resources, but hardly suitable for modern browsing needs. For example the JavaScript support is still in its infancy, it has no support for HTML5 video, etc. On NG it would be a huge step backwards even compared the outdated Odyssey 1.23.
-
That's why I'm thinking to make one team beyond division to make one browser for all. Than community can support team through bounty or donations every year. Of course we must ask developers to join team for example deadwood, Fab, kas1e? Or you if you want?
If not Odyssey what other browser could be ported to NG systems?
-
It is step backwards, thats true. Just wondering these endian issues, if webkit team is not going to fix it and they do have a attitude that it doesn't matter? Does it mean that these problems are increasing in future? These kind of projects tends to be single person projects in amiga land. ?
-
What a damn shame. PPC had an amazing browser on MOS and its dead in the water now because of silly things like this. What about a 68k solution? Same issue?
not long ago olaf approached you guys on morphzone with his usual ideas of cooperation. what he has earned afair was claims that morphos doesnt need anything from anyone in the "amiga" scene, including aros, all traces of aros code in morphos have been rewritten long ago and that only other parties might gain advantages from such agreements.
when deadwood was trying to discuss bounty for a task, none else is apparently intending to address, namely the said endian fix, none of morphos or os4 users have spoken up. fine. now hire some vietnamese coder, if you like..
-
It is step backwards, thats true. Just wondering these endian issues, if webkit team is not going to fix it and they do have a attitude that it doesn't matter? Does it mean that these problems are increasing in future? These kind of projects tends to be single person projects in amiga land. ?
May I ask why have you given up on Odyssey so easily? :confused: If you can't even find someone to fix JavaScriptCore, do you think you'll find someone to develop NetSurf into a modern browser, which is at least an order of magnitude more work? What you are saying is the equivalent of buying a new house, because the window previous one got broken.
-
Netsurf is great for platforms with limited resources, but hardly suitable for modern browsing needs. For example the JavaScript support is still in its infancy, it has no support for HTML5 video, etc. On NG it would be a huge step backwards even compared the outdated Odyssey 1.23.
I think utri was suggesting some people could help to implement/improve these features.
They are being dealt with, but it's a small team so progress is slow. Patches are usually welcomed! http://www.netsurf-browser.org/developers/
edit and of course working on NetSurf core doesn't need an Amiga developer nor a PPC machine.
-
Fab did and was told iirc "we don't even own a ppc machine so you'll need to do it yourself"
We maybe should do something like that "What do you need from us to fix the indian problem..blabla..." I suppose that Linux PPC could be run under a VM to test the build...A diff between the last working version vs the first to break compatibility might shed some lights to fix the problem in an easier way.
Kamelito
-
Now I need to choose my word carefully. :) There ia a team working with netsurf, pushing it up to date. Being part that team, might be less time consuming and require less work that fixing Odyssey at least from single person point of view NOT for generally speaking. As I said these kind of projects tends to be single person projects in Amiga land. There should be team for fixing this, but .. who knows? And after fixing this problem, what happen next if webkit dev team doesn't care a endian issues? What will be next issue with it?
-
Stupid question, but how we can find someone? Inside our community or better outside?
Or maybe we can find other alternative for WebKit.
Google's WebKit fork Blink and its V8 JavaScript engine works on Power ISA and is actively maintained by IBM.
-
May I ask why have you given up on Odyssey so easily? :confused: If you can't even find someone to fix JavaScriptCore, do you think you'll find someone to develop NetSurf into a modern browser, which is at least an order of magnitude more work? What you are saying is the equivalent of buying a new house, because the window previous one got broken.
http://sourceforge.net/projects/leopard-webkit/
-
Google's WebKit fork Blink and its V8 JavaScript engine works on Power ISA and is actively maintained by IBM.
If this is better solution that Odyssey why not? If we start on the base that is supporting PowerPC it will be easier for us to have updated browser for NG system. I think there should be some options, Linux PowerPC community have some modern browsers on their systems? And when it is possible to develop one browser for all NG systems it will better. We need cooperation in this case.
-
If this is better solution that Odyssey why not? If we start on the base that is supporting PowerPC it will be easier for us to have updated browser for NG system. I think there should be some options, Linux PowerPC community have some modern browsers on their systems? And when it is possible to develop one browser for all NG systems it will better. We need cooperation in this case.
Can someone familiar with the Odyssey codebase tell us whether adapting Odyssey to use Blink/V8 instead of WebKit engine would be viable? Or have the two engines diverged too much these days? Fab? Deadwood?
-
I propose this for the sake of ending the arguments:
Step 1, appoint someone trustworthy and knowledgable to dig into this problem. Perfarbly someone who is pro all camps.
Step 2, find all tangable solutions (fix component, use alternative etc)
Step 3, find someone inside or outside the community with the time and talent to fix and maintain component (and not take forever doing so)
Step 4, ask if a bounty will do. If yes then
Step 5, open up a bounty
If it's Fab, Deadwood and/or Kas1e, all the better.
My 2 cents.
-
I have asked WebKit developers on the Twitter @webkit and they abandoned and will not support PPC any more.
So it looks like or we fix it in our community or in this situation I think is better to jump into new engine? Because it could possibly that after one fix WebKit , then in some time we must make other one etc.
-
I have asked WebKit developers on the Twitter @webkit and they abandoned and will not support PPC any more.
So it looks like or we fix it in our community or in this situation I think is better to jump into new engine? Because it could possibly that after one fix WebKit , then in some time we must make other one etc.
wasnt that what you have been told from the start?
btw. on morphzone bigfoot reports to have fixed most endian problems with webkit engine in order to fulfil the js jit bounty he has accepted once, organized by pampers. the question is, what version of webkit engine it is, he fixed, and how could that remain maintained, since the fix sounds like a lot of work and doing that every other webkit release woulnt make any sense. imho the maintability is the first question people needs to ask themselves how to solve, especially if webkt team wouldnt accept big endian patches ushed upstream to their main repo, which seems to be the case.
once this is considered, bigfoot could be paid off, submitting his code to the deadwoods repo and the remaining issues could be taken care of in ccoperative manner. the question is, if morphos users who backed up the bounty so far, want to cooperate in such an affair, even though the resulting source was likely to be released anyway.