Quote:

I wholeheartedly agree. If we can get any increase in performance by some new graphic drivers I’m all up for that.

But would you be prepared to pay for more modern drivers like AmigaOS 4 users have to? I probably would but not sure that would be the case for everyone.

> there are other aspects of web browsing in 2021 that need
> some muscles other than video playback. The PPC is ancient
> in performance [...]. The Macs we use are at 2005 level at best.

Agreed. Nonetheless, I really would like to see how a 2005-level PPC machine with a 2010+-level graphics card and a good JavaScript JIT compiler would cope with modern websites involving hardware-accelerated video decoding under MorphOS :-)

> so at that point

I think we may be talking about slightly different points in time. I'm talking more about some kind of early preview, suited to whetting the appetite of existing MorphOS users. I'm aware that going by the release history of the MorphOS team, the chances of such potentially unpolished release are extremely slim.

> why not just [...] write some drivers for some motherboard [...]?

Driver development is difficult, you know :-)

Indeed, and I agree, but there are other aspects of web browsing in 2021 that need some muscles other than video playback. The PPC is ancient in performance compared to even the cheapest low-end computers of today. The Macs we use are at 2005 level at best. Sometimes this becomes quite obvious, like when browsing the heavier web sites of today…

>> We need faster CPU’s. There is no way around that fact.

> ;-)

...or better use of compatible hardware, like hardware-accelerated video decoding. This of course would only work for exchangeable GPUs, so Mac mini etc. are out of the game.

@Andreas_Wolf

Yes I understand your point.

My point was just that if you have already come so far as having a full-blown 64-bit OS with SMP support, then you have a system that’s already not even source code compatible the old world, and most of the heavy lifting with development should already be done as well, so at that point - why not just set the compiler to make binaries for the new architecture and write some drivers for some motherboard and complete the transition?

But at this point I’m doubting this as well will ever happen…

>> As an interim step [...], there is still the remote possibility [...]

> why hold back instead of going all the way when you are at it?

Doing an interim step doesn't necessarily mean holding back. Sometimes, interim steps make sense when the final product isn't quite ready yet. I think I would at least try a 64-bit and SMP-enabled MorphOS on my multi-GiB-RAM and multi-CPU G5 machine before considering the purchase of a MorphOS-compatible x86-64 machine. I don't expect this to happen, though (hence "remote possibility"). Of course, such decision would depend on the required effort, and I'm well aware there's more to having an OS run on a certain architecture than just the kernel.
In the end, the MorphOS team will have to determine what makes the most sense to them. I'm confident that whatever decision they arrive at, it will be in the best interest of their OS :-)

@Andreas_Wolf

If you break the "Amiga" in such a fundamental way by going 64-bit, then why hold back instead of going all the way when you are at it?

@NewSense

Quote:

;-)

[ Edited by takemehomegrandma 02.07.2021 - 15:57 ]

> Many, and probably most, MorphOS users (probably less than 50%)

;-)

> The next step of MorphOS to 64 bit hardware has not been
> announced, so for the forseeable future the PPC hardware
> is what we need to be able to get the most from

As an interim step to MorphOS for x86-64/AMD64, there is still the remote possibility of MorphOS for 64-bit PPC hardware, i.e. MorphOS running on G5 Macs and the X5000 in 64-bit mode. After all, the new MorphOS 64-bit SMP-capable kernel was developed from scratch on and for the X5000 before being ported to x86-64/AMD64.

https://morph.zone/modules/newbb_plus/viewtopic.php?topic_id=12995&forum=11

Quote:
I take your point on-board, which is basically a fair comment that most systems with such hardware have been ditched.

That said, MorphOS doesn't function like most modern OS's so it's somewhat out-of-step to compare it with other OS's that require more "grunt" CPU power to get the same speed/functionality per cycle, but tying a modern WebKit engine into MorphOS is making that process more comparable I have to admit, but it's the only way to keep MorphOS's web-browser as up-to-date and relevant as possible. It's the last Amiga-like web-browser that is still basically capable of accessing most of the internet as it is these days.

My current concern though is based on errors/discrepancies still needing to be fixed, that of RAM being "swallowed" when it shouldn't, and that audio both in videos and as standalone audio, is not being processed when links are clicked on, unless you use some abstract way of getting it to trigger - that should not be required, and it should all work correctly, with the RAM not disappearing in front of your very eyes. Videos on many websites you go on can still respond a bit flaky, though much of what Wayfarer can handle is truly amazing.

Yet something(s) are not working correctly IMHO if that is happeneing and surely seems worthy of further investigation and where necessary fixing. So, let's not give up on what may well still be achieveable.

Many, and probably most, MorphOS users (probably less than 50%) won't have a PowerMac G5 with a CPU/Dual CPU higher than 2GHz so Wayfarer's development would get under-used and stifled by not being able to handle virtually all videos on social media and even such places as YouTube. The next step of MorphOS to 64 bit hardware has not been announced, so for the forseeable future the PPC hardware is what we need to be able to get the most from - that's my point.

The step-change of the internet of the last few years has slowed-down what used to be readily accessible websites that hosted such videos, as I obviously remember being able to view, or at least download, videos that were hosted on such platforms that were readily playable within the capabilities of our mid-range "extinct", but useable, Apple hardware with MorphOS - which they are now struggling to access nevermind play, and certainly for the most-part can no longer download.

I'd like to see what else can be achieved with our "ancient" hardware, by Jacek and whoever else is involved in making Wayfarer capable of what it has already amazed us with - I do really mean that - and not just give-up in these comparatively early stages of development. I'd just like to see some more Jacek "magic" . . . who wouldn't?

Quote:
C'mon dude, I hope you realize it's more or less a miracle (as in a tedious ammount of work to do) that you can watch 480-720p online media in your browser at all on 17 year old G4 hardware.

It's not about simply porting an existing browser engine to MorphOS and be done with it. A tremendous ammount of work must have been put at optimizations on the MorphOS-side of things on all fronts, even adding parts to or restructuring parts of MorphOS itself! I can also imagine doing more optimizations has the side effect of diverging from upstream webkit which makes porting new stuff even harder.

To get an impression on how upstream webkit performs on your G4 just try Epiphany on Void Linux. It won't do video playback at all. Or look for any other browser performing that well and that fast with todays websites on compareable PC hardware from 2004 (which means single core 2-3 Ghz Pentium 4). You won't find any. And I actually tried on my disposable 3 Ghz P4 HT which I keep for Linux testing purposes...

@ jacadcaps - thanks for your fast reply, and offers to improve "the impossible", but here's an interesting factor . . .

I managed to get the podcast to play through on the F1's website, by a bit of tab-foolery, jiggery-pokery.

So, I loaded up the F1's Christian Horner podcast webpage, and clicked the link/load button (>) and nothing happened, but I also loaded the additional (same) podcast that is also hosted on audioboom.com that I mentioned . . . by then clicking on the weblink for the audioboom site, and then going back to the other tab for the F1 site, after a couple of clicks on either/each one-at-a-time the audio then started on the F1's website link, but if you pause it for a moment then you have to go through the whole process again to get it to start playing again, but it does play, but obviously it shouldn't have to be made to work like that.

So, by that process, surely it shows that there is something that needs fixing in the way audio is being handled. I tried clicking solely on the F1 website link time-and-time again, but that never triggered the audio to play, only when I clicked on the link for the audioboom podcast to play (which never played) did the F1's website link trigger the audio to play the podcast, so it can be made to work, but surely it should work without all that messing about, and so surely that is nothing to do with CPU capabilities just getting Wayfarer to react in the required way to start playing the link, by whatever means it requires to do so?

Quote:
OK, I see your point, but it's not because I am so stubborn, it's because I'm not developing Wayfarer so I don't get to see the management of resources like you do as a general rule.

However, it seems like there is CPU capability to just about manage the site within the hardware resources, but some of it is running amok, and not doing what it should (RAM misused, audio not playing, JavaScript stop-start-stopping video from playing). I noticed that the audio thread is only using ~2% of the CPU, though other Wayfarer components, as you rightly point out are using a far larger proprtion of the CPU/resources, but that doesn't explain why RAM is being "gobbled-up" for no apparent purpose that is actually audbible or viewable within Wayfarer - it is either mishandling some of the data being thrust on it or something isn't working right - even if it is having to work hard to handle the data.
Quote:
So what speed CPU do you consider is required?:-?

Many MorphOS users will have a 1.25GHz Mac Mini. I've got a 1.5GHz Mac Mini, but even with a PowerBook 1.67GHz that's little more CPU grunt or RAM to handle the data. So, if my hardware is 700MHz short of what you consider to be necessary, that makes a CPU of 2.2GHz to be the absolute minimum requirement, I have a 2.3GHz PowerMac, but that would then be the G5 and there are not that many MorphOS users who probably own one, or I would imagine, want to be running their power hungry PowerMac G5s purely to view certain webpages. I find the MacMini capable of most of what I need, and I feel with a few more tweaks and fixes in place, which might require some better control of handling requests from websites that it could be made to work more effectively even on these modern websites - at least I hope so, and I feel sure you will do all you can to make this as possible as could reasonably be expected.

Blocking JavaScript will briefly improve video playback since that's done on another thread(s), so you get a short CPU boost. Short because JavaScript is needed to download video/audio data in first place. It's also needed to release the RAM. A mac Mini is TOO SLOW for this website, period.

Since you're so stubborn, here's some links for you to watch and compare:

https://youtu.be/Iv38R8l3EHU G5 PCIe 2.3GHz
https://youtu.be/T2z54XSRo3w mac Mini 1.5GHz
https://youtu.be/pqce-7x1e24 same mini with focus on CPU usage

That said, I've managed to make some improvements, but there's only so much I can do when your CPU is basically 700MHz short.


[ Edited by jacadcaps 01.07.2021 - 04:13 ]

@ jacadcaps - thanks for your quick reply, much appreciated, but ...
Quote:
OK, but why will the video play correctly then, as surely the JavaScript should be handled correctly inside the browser's mechanisms, which it somehow seems to be misbehaving currently and not doing what it should. In fact when I do get a video playing it seems to miss off the first few seconds of the video file and play - part-way along, but with no sound (I have amended my AHI prefs to what you have suggested - but with little, if any, benefit/success) for most occasions I use it/try to play a video, but the BBC site handles videos/movies/sound it in a well behaved manner and it all works as it should.
Quote:
Bear in mind that RAM is consumed within seconds, so even if I had 830MB RAM more then that would be "gobbled up" within just a few seconds it seems. I've watched the RAM disappear, it goes in chunks of 100-200MB within a moment (1-2 seconds), I don't think your"solution" is a realistically workable idea to the obvious problem it is. It seems to me that some memory handling process internally in Wayfarer is "running amok" - out of control on the Formula 1 website, in this regard.
Quote:
So, having changed the AHI settings to what you have suggested has not improved the situation - still no sound, and that's if I "catch" Wayfarer before it "gorges" on whatever RAM I have, usually 500+MB or so, within a few seconds - and then crashes.

There's currently a PodCast with Christian Horner on the Formula 1 website, and obviously that doesn't require video capabilities, but even with the suggestion of changing AHI prefs, which I have complied with, there is no audio, but the PodCast seems to be downloading within the Wayfarer data stream, but is just not getting played, as my RAM resources are dwindling slowly but surely - with no benefit of hearing what was recorded in the podcast.

As soon as I release/close that page/tab the RAM resource goes back to where it was before I loaded the page and clicked on the podcast, so something is definitely not working as we could expect it to. I started with about 400MB and now it's down to 326MB and continuing to drop, so that's 74MB for a podcast to play - which doesn't, but is continuing to "hog" Wayfarer's resources. As I've typed this remark the RAM is now down to 301MB, and the download activity at last seems to have completed, so I might assume the podcast is about 100MB of data in size(?) - but still no audio at all, and the RAM resource is still dropping and I'm doing nothing extra to incur that penalty/change, but I can load WebRadio or play an MP3 or a CD or any other audio, just not get any audio from the Formula 1 website.

Additional to that, the same podcast is also hosted on https://audioboom.com - Christian Horner on Verstappen's evolution, RedBull's bold ambitions & Vettel v Webber . . . and I cannot get the audio to play from that website either, so surely it is something to do with how Wayfarer/MorphOS handles audio that is affecting this issue?

The only other thing to mention is that on the audioboom website my RAM resources are barely affected.

Quote:

Drop down is a blocking menu so you are effectively preventing JavaScript running on the site while you keep it open.

Quote:

Configuring AHI to use device.audio should help. Set Music Unit to Unit 0: HIFI 16 bit stereo++. As for memory consumption I can only suggest you try it on a machine with more RAM.

Quote:

The reason is misconfigured AHI. f1.com tries to play 2+ streams at once to automatically manage the quality (ha ha) which is the source of your issues.

I've tried to get some improvement(s) with Formula 1's website, but generally with Wayfarer v2.2 I don't seem to be able to find a "sweet-spot".

Sometimes I can get a video to play (but usually no audio) most of the way through, using the same attributes:-

Wayfarer (as Wayfarer),
and with these options active (ticked)
Media Playback
HLS Streaming Support
Decode Video

I even found, in some instances, that if I kept the drop-down menu in view then a video would play (which seems weird) when otherwise it just kept start/pause/restart/pause, etc. (in a seemingly perpetual loop) and failed to play, but with the menu on view it played until the looped/streamed section of the video was used up, and then I got the "unable to download video" error message.

Sometimes the audio starts, but there is no rhyme or reason as to why it starts some times and other times fails to load the audio. It also gobbles up the RAM and often/usually plays part way through a video and then exits with the "Wayfarer seems to have run out of RAM" error message/crash - even if the video is only 4 to 5 mins in length which seems to be an "out-of-control" situation with Wayfarer. After the error/crash I have to reboot as there is no way Wayfarer will load audio for Formula 1's website or work correctly as the "CON" and "Wayfarer" are Frozen tasks in the system and only rebooting will free them up again.

Some aspects/uses of Wayfarer on a restart/launch - without a system reboot - will work, at least for a while, without issue(s) other than audio on Formula 1's website, but then, audio working on Formula 1's website is sporadic to say the least, for no apparent reason.

I have noticed that with v2.2 there is a major improvement with video, with audio on the BBC iPlayer website, so I watched quite a long section (10 to 15 minutes) of "Antman" - one of the TV stations movies of recent days, play via the iPlayer, and it seemed to be pauseable, playable/restartable from a pause, with audio sync'd with the video, and would play nicely in full screen mode even on my MacMini G4 1.5GHz 64MB, so that seems to be a step forward.

Yet another step forward for Wayfarer - I have noticed is that it is now possible to access the "Sounds" webpage of the BBC, so you can listen directly to the BBC website's radio, podcasts, . . .
and anything else that is broadcast via their online services.

I also noticed that the right click context menu on eBay's message pages is now working again, which it hasn't been for a few releases of recent.

> using Wayfarer - as Wayfarer (no User Agent)

Wayfarer does emit a default user agent string. Technically, it is possible to have the browser emit an empty string (i.e. no user agent), but I'm not sure how websites will react to that.

Quote:

The shutdown problems with all my 3 systems are gone with this release.

Thanks Jaca.

Quote:

Any ideas regarding this issue? Downloading archives from Bandcamp still doesn't work in 2.2, no matter which format I choose. Nevertheless, great work! Version 2.x is rock stable here.

[ Edited by analogkid 27.06.2021 - 09:14 ]

Quote:
That's great, but I've already had a few crashes - 1 no Crash Reporter, 2nd & 3rd - Crash Reports sent - after at least a couple of the crashes the WkWebView Acinerella Video Decoder had gone "Frozen" so I had to fully warm reboot to be able to use the Formula 1's website to test the functionality as I obviously couldn't use the video capabilities with Acinerella "Frozen".
Quote:
sadly I've not found that to be working with Formula 1's website - videos can play, but then after about 16 seconds they loop/repeat whether the video is on SD or HD or Auto
Quote:
I trust they all work - as I've not had direct occasion to test these fixes.
Quote:
If only that were true, but there are still issues.

The best I got on Formula 1's website to view videos with v2.2 was using Wayfarer - as Wayfarer (no User Agent),
and with these options active (ticked)
Media Playback
Media Source Playback
HVC Codec Support
Decode Video

Then a video started with audio and video - in sync, but only usually played for about 16 seconds then it stalled (video) and then looped the audio first, then restarted the video all by itself sometmes and looped over and over again those same 16 seconds of video/audio, with many, same 8-16 second sections, stalled in between, though clicking on the video twice (pause || play >) often restarted the video/audio but only back to about 8 seconds from the start and then only played until about 16 seconds and then looped again. Sometimes the video played for about 1 minute and 30+ seconds, then looped back on itself and repeated what it had played before not wholly but most of it, then it eventually moved forward a bit, onto 2+ minutes then dropped back to the 1 minute section and played that all over again.

The video cache seems to cope with the video data, but video data downloading stalls while the video does its backwards/forwards/backwards movements, but gradually shuffles forward bit-by-bit and begins downloading when it gets past the point the website will offer more data. I should also mention that as the cache is filling it is holding more and more data in RAM and so for the 7 minute video as it gets to about 2/3rds of the loaded video it has "swallowed" about 400-500MB of RAM, so I got close to "bottoming-out" of RAM and of Wayfarer crashing again.

Surely a webpage of a 7 minute video playing at SD quality shouldn't be holding 400-500MB of RAM captive?

So as the video moved on a bit further and then repeated what it had just played before, like a video echo. Don't forget the video I refer to is supposed to be 7 minutes and 3 seconds of playtime, but I'm only seeing about 20+ seconds of that video at a time, before it first "peats" then "re-peats" and I'm getting "tortured" during that time span of 16+ seconds - it's like being in a déja vu video nightmare!

2.2 is now out.

- Fixed duration handling in the Media Source player
- Disabled focus frame on some elements since it wasn't always redrawn correctly
- Fixed a low-memory related crash in the PNG decoder
- Tweaked ICU data checks to try and detect broken installations
- Added a workaround for Wayfarer not shutting down (may still be up to 30s)
- Fixed setting media playback variables after prefs were reset

@NewSense: formula1 should now be more or less OK :)

[ Edited by jacadcaps 27.06.2021 - 02:37 ]

I am aware of the Formula1 issue, happens on YT too if you change quality (same thing as F1 changes quality for you). Not sure when I'll find the time to look into this. I suggest you use 1.21 for that website for the time being.

I downloaded v2.1 and installed it as suggested with all the new files in the download placed in my system with all the unarchived folders/files.

However, I have had a couple of crashes when trying to view videos on Formula 1's website - without HLS active, though videos only seemed to play for 0:08s anyway with video/audio, and after 0:08s only with video. Initially, on that use/video, it just crashed out, and I got a Crash Reporter which I completed/sent.

The second time I loaded a video from Formula 1's website the video played (no audio/sound) and I got an error message from the website - this time with HLS active, stating "Could not download the video, Error Code: PLAYER_ERR_TIMEOUT, Session ID: 2021-06-23:aa5d671bb15b070e9a0d1da2 Player Element ID: 1703352181737740921, but after that error message was clicked on "OK" it then continued to play the video - though still none of the audio worked during the video playing, but then it crashed out stating that "Wayferer seems to have ran out of RAM" or words to that effect - but no Crash Reporter to complete.

The sub-menu mouse middle-wheel scrolling does seem to be working as it was in earlier versions, so that's definitely an improvement from v2.0.

@ jacadcaps - I'd really like to see the "Save As" download option implemented as soon as you can, as this is a real "pain-in-the-butt" only being able to save a file with the name supplied from wherever it is download from, and only to the path that is "set in stone" in the Settings. I really would appreciate this option being part of Wayfarer's Settings as it really is a nuisance not having this dynamic user choice.

Thanks Jacadcaps for this new update!

Fantastic work!