Quote:
Might that lead to web-browser "Font Envy"

Quote:
Font fingerprinting

Hmm ... why not (also) debug the other way round. Try to do things which you suspect should make a currently working browser (MiniBrowser) stop working. Like disable JIT. Run on slow emulator. Or single core machine.

Site on MiniBrowser still worked after 'export JavaScriptCoreUseJIT="0"'

Sometimes info you get from debugging can be deceiving. If it has to do with timing, maybe it's not the timing itself, but timing code implementation/precision.

From the debugging of those issues I've done so far I can tell you that those checks rely on timings as one of the markers. That's something that will obviously be totally off on a single core with no JIT.

webkitgtk-2.32.3's example MiniBrowser on Linux x64 at least seems to access the site mentioned by NewSense (https://www.zolo.ca/) just fine.

Do you think it would still fail on little endian 64 bit MorphOS? Why? How can the site know that it is not being accessed from Linux machine but some other OS, if you don't tell it?

@NewSense

Quote:
Just to be clear, it will never be achievable.

Cloudflare's browser check is more complicated than you assume it is.

More specifically, Cloudflare is a not a "one-solution-fits-all" service. You can sign up for a free account and take a look yourself. Website owners are free to choose what should be checked if someone attempts to visit their website.

Just because you can visit one website protected by Cloudflare, this does not mean you can expect to visit all of them. For instance, some website owners may choose to block specific user agents.

In general, it is also fairly trivial to detect if a user agent matches a browser's underlying engine. Since Odyssey and WF use Webkit, choosing "Safari" as a user agent is certainly your best bet. If you choose "Firefox" as your user agent but run a Webkit-based browser, being deemed a "suspicious visitor" is not surprising. (The same applies to pretending you run a Blink-based browser such as Chrome or Edge.)

@ Piru & jacadcaps - OK, I'll leave it there, I think you've offered fair insights into the matter, and I think if you thought there turns out to be a way in the future to "fix" this issue then you'll try your best to do so - as usual.

So, I've got replies from you both, and you've been very fair in corroborating your replies, which I really appreciate, and I don't want to be seen to be annoying you/pestering you on a topic that isn't achieveable currently.

Quote:
The problem is likely way too complicated that it could be figured out from differences such as these. There likely are multiple factors, and they might have some interactions between them that also are not obvious.

As mentioned earlier, the page in question is also deliberately trying to hide what it does by being highly obfuscated. This makes it really frustrating to try debug the problem.

Finally, the service in question is likely also constantly updated (to combat attempts to bypass the protection). Hence any work you do on it might be affected by these changes. Even if you would figure out what the problem currently is and fix it, the page could suddenly change and break again.

All these factors make it unlikely that this issue would get prioritized.

@ jacadcaps - OK, I get the situation you're in/explaining.

However, can you second guess what such software would be checking for with a specific web-browser or feature that it offers openly to another web server for it to allow that person access to such a website, i.e. what does Firefox, Chrome or any other mainstream web-browsers offer that is likely to satisfy such checking, that Wayfarer doesn't offer generally when using the Wayfarer, Chrome or Safari User Agent - that when using the Safari/iOS User Agent it offers that website details that allow it to connect through?

That's so Wayfarer can incorporate such features to overcome this issue, if you think it could be possible to make such an intuitive, logic-based "guess" to improve Wayfarer with more User Agent capabilities that work more widely and effectively.

I ask because I got through the check, after many attempts, by using the Chrome User Agent, and then swapped/reloaded the page as Safari/iOS which seemed to do the trick (complete the webpage/Cloudflare server check). So, surely there is something in what Wayfarer forwards to a website - when these User Agents are chosen that indicates it's a "friendly"/suitable web-browser to these sites in question, that the other options don't offer that might make the other User Agents more effective straight away if implemented in them all?

As I have previously mentioned, I have no way of debugging code that was deliberately designed to make debugging it as hard as possible.

@ jacadcaps - do you think you will be able to get Wayfarer to access websites that have Cloudflare security on websites - as they seem to be [generally] inaccessible currently to Wayfarer as it cannot override / understand the protection preventing it from accessing them, as you get to the hCaptcha check, which you can complete, but then it "bounces" you back to hCaptcha again, and again, so you get nowhere - such as the site link I have passed above.

[Update] - I tried all the User Agents previously, but I now find that using the Safari/iOS option after Chrome did allow me to get access to that site - whether it works with all sites I don't know, but I have had issues with accessing other sites that have "Cloudflare" website protection.

Quote:

@Jaca : you rulez! Thx!

https://www.youtube.com/watch?v=rL5cTwdOtQs

@BatteMan: your stuff loads now :)

[ Edited by jacadcaps 23.07.2021 - 03:06 ]

@jacadcaps: thx. While waiting, I'll use Wayfarer to edit and OWB to test my IF.

Quote:

Just something I have not bothered to implement. Maybe later.

@BatteMan:
The website most likely fails because Wayfarer only mocks WebAudio support (it's there, but doesn't do anything). Not sure when I'll have a chance to look into this.

Don't post here since a long time but I want to say it again : thx Jaca, Wayfarer is a really incredible and awesome piece of software.

I want to point out a "bug" with games (Interactive Fiction) created with Moiki.fr. If you try to launch a game, the loading halts with the downloading of sounds ("téléchargement des sons"). You can try one of my IF here : https://moiki.fr/social-club/batteman/speleologie-en-aveugle/play (it's in french, sorry ^^)
(OWB succeeds to load it, after a more or less long time).

But, the very good news is we can use Moiki Editor with Wayfarer !
So, we can create and edit IF ! But we can't test them, even in the editor, for the moment.

Thx !

This is a link to a Google translation:
https://translate.google.com/translate?sl=auto&tl=de&u=https://www.amigalove.com/viewtopic.php?f%3D7%26t%3D1911

As you can see - or at least here, in my Wayfarer 2.3 - on the right side there is a scrollbar but ist does not show that I could scroll down. I can scroll down but not by using the scrollbar...Spoofing as Chrome you could assume that it should work...;) So, it is a bug?

Ohhhh, just a reason more to get another one...;) Thanks for the hint!

In this case I only had MorphOS Storage and moph.zone opened but maybe it is also due to the graphics card...?

Your graphics card isn't supported for video playback. Radeon HD 5450 has no hardware overlay support. Please see https://morphos-team.net/hardware for a list of supported graphics cards.

Quote:

Is it the only tab that's opened?

Unfortunately I do not get Wayfarer playing YT videos...:(
Yes, I have copied the cgxvideo.library to MOSSYS:Libs.
Example: https://www.morphos-storage.net/?id=1652577
I can see the start screen ("Broderbund Softare presents..."), if I then click play then the white circle starts turning but the background stays black...

My settings: Media playback enabled, spoofing as Chrome, enabled all types of codecs....
Any idea? In OWB it works...Thanks!

PS: By the way, it feels like that Wayfarer becomes slower. Even MorphOS Storage takes some seconds...

@ jacadcaps - but what is the Required User Gesture?

Though I see the v2.3 menu has now been amended to state "Requires User Gesture to Play" which explains it better, and I see the MUI help seems to have a better explanation - "Block auto play of media without user interaction"

Would "user action required to start to play media", maybe with "allows website to load all initial media data prior to user triggering playing" explain the meaning of the option button more simply? Though your explanation above makes understanding why the "gesture/action" should be implemented for certain sites, and is in effect probably a last resort to allow the full website media data that is being requested to be loaded - has to be delayed until the full initial loading has completed before a user takes specific action to start the media playing.

This was not clear at all, well not to me, to understand why the "Requires User Gesture ( to Play v2.3)" radio button was there from the initial release, and in what circumstances it should be considered or triggered as a last resort, but I understand its use now - following your explanation.

Requiring a user gesture basically avoids player starting to play before the website is actually loaded. Things can easily go south with many JS players in that case, due to timings. Better to wait until the website loads, then start playback.

Quote:
I've not had a problem that I'm aware of with the USB2 connected SATA HDD - but obviously as/when I've had "crashes" with Wayfarer then maybe this affected a file/part of a file in the cache, but so far all has been great with v2.3 !

I don't intend to put Wayfarer on my main system HDD as yet, but I feel it's almost time to do so, and I take on board yours and others comments about using USB2 for such intensive small file(s) transfer/loading.
Quote:
Most of those settings were as commented above, but I had Spoofing/User Agent set as Safari - why select Wayfarer?

Also, that option for Require User Gesture was OFF - but why must "Require User Gesture: ON? As I don't see any User Gesture required as such before it starts to play the video, please explain why this is so important?

Frankly I would never have thought to have the Require User Gesture set to on, as I would never have thought that was important for audio to work, or for whatever it achieves. As I never realised there was any "gesture" I needed to take or action I took to make audio get triggered, even now when I do get audio playing on the Formula 1 videos - thanks to your great options/advice.

Just a query regarding other sites that have video that will require audio output - should all such sites, i.e. YouTube, have the Require User Gesture - set to ON (just in case that setting is as important as it is on Formula 1), and should the User Agent be set to Wayfarer, or any other settings that are necessary for such sites.

Ideally a better breakdown of what the menu/settings options do to trigger video/audio codecs on such webpages would probably help understand what is needed to be set on/off for a webpage by webpage basis.
Quote:
I find that even with the USB2 SATA HDD the data is spooled quickly enough, and the videos play through almost seamlessly, with audio - within the limitations I expected (SD video), but I'll bear in mind the advice about using Wayfarer from the system IDE channel in the future, but for now it's entirely acceptable - thanks.
Quote:
For me, as long as I know which options to choose to make it work then I'm happy with that, it was just the obvious disappointment of watching a video with no audio - that's now solved - which is great, so for me v2.3, along with your great advice/settings, is a good step forward.

Quote:

As others have pointed out, it's most likely a USB issue. My guess would be some http cached file got only partially written to drive due to crashing/rebooting/usb/whatever and you were being served half a JavaScript file instead of a full one, or something like this. I have this happen once in a while but I crash a lot when doing some of the tests so it's kind of expected.

Quote:

Loaded https://www.formula1.com/en/latest/video.race-highlights-2021-austrian-grand-prix.1704370404771962147.html just now on a 1.5GHz mini. Waited for page to load (until CPU goes down), then Settings menu > www.formula1.com >
- Media Playback: on
- Media Source Playback: on
- Require User Gesture: on (that's important)
- HLS Streaming: on
- VP9: off (you really don't want VP9 OFF a mini, it's way too slow)
- HVC: off (it's also way too slow on a mini)
- Decode video: on
- Spoofing: Wayfarer

If "Require User Gesture.." was off before, you might want to reload the page now. Once ready, press the big play button, wait for a bit, it's start loading and show you the preview picture again with the player bar accessible. Select SD quality, then press play. It's not gonna be great but it will play, with audio. The player on this website alternates streams mid-play (changes qualities as it sees fit), so you'll see some stalls. Nothing I can do about this, this is fully controlled via JavaScript.

If it's still too slow, put Wayfarer on a faster drive. Loading websites writes a LOT of data to a drive and using a slow medium like USB will negatively impact performance at a time when you already have no margin left, if only because USB I/O normally needs more CPU than regular HDD I/O.

If it's still too slow, tough luck.

I'll look into why there's no sound when spoofing as Safari/iOS when I have the time.

Edit: after a quick analysis of the HLS stream (the one you get with Safari/iOS), turns out this is a special case (oh joy) where the audio is a separate HLS sub-stream. No promises on whether this is ever going to be supported, as it's a lot of extra work for a single website.

[ Edited by jacadcaps 08.07.2021 - 15:56 ]

Quote:

We have exFAT on MorphOS, but this does not fix the USB issue.

You would not use a hard drive that stops spinning and restarts every now and then. Same happens to USB, it is just called a NAK timeout.

USB should be used for data transfer or installing purpose, but not for running heavy duty operations that constantly read and write megabytes of data.

The USB protocol (at least USB1 and USB2) is just bullshit and a result of keeping controller chips cheap. Unlike firewire, which is way faster on equal bus speeds, you need to do most stuff in software. The first iteration even required 32 bytes overhead to transfer 32 bytes of data.