@xenic

LocalStorage was introduced with HTML5. To be simple, it basically does the same as cookies and stores data in a (sql) database.
Many sites use localstorage nowadays, but you can still disable it globally or locally in Odyssey settings.

@LiveForIt

I use YV12 anyway, in overlay mode. It makes an even greater difference.

@tlosm

The HTML5 player in Odyssey for MorphOS is actually about as fast as MPlayer (modulo network overhead) when running in overlay mode.

So on a PBook G4, 360p eats about 30% CPU, and on a G5, 1080p can be achieved smoothly in overlay mode (and 720p is fine in window mode).

@LiveForIt

Mouse wheel support is fully functional in MorphOS sources, and it was not removed or anything.

At some point, the wheel actions were hardcoded in cgx stuff file, but i moved that to the generic input handling instead. So currently (and in the available sources as well), i just forward the wheel event to the mplayer event handling.
You can then decide in input.conf what the wheel should do. It's much better than hardcoding it.

The problem is URLOpen is really badly designed, and not having REXX support is just unacceptable. If URLOpen wasn't so limited, the choice would be easy.

@K-L

Not really. It very much depends on the codec, some were sped up a great deal too. Regarding H264, i regularly compare it against a test video, and it's only marginally slower in the latest version i released on MorphOS (2% or so...).

@LiveForIt

The MUI GUI code is not a mess at least, it's handled in gui/morphos/ directory along with the other toolkits...

cgx_overlay_gui has some relations to the gui system, so it has to support some specificities, but it's very minimal, actually.

@Raziel

I used ahi library (or lowlevel) api for a good reason, because it had less latency.

kas1e won't be able to change that easily, it's quite some work to change from library api to the device api (which is more limited).

For the record, audio IS released when paused.

Anyway, there's a good workaround: just use device.audio as music unit in AHI settings. That way, programs using music unit will be redirect to shared units.

@sofistisoftware

Odyssey doesn't generate any env var at all for that, it deals with it internally. I just let the opportunity to override the user agent with an env var too, if someone really needs it, but that's all. And keep in mind that this env var overrides everything, including per-url settings, so i suggest not using it unless it's really needed for a good reason.

@ChrisH

P96 not working in the amigfx driver is not a bug. It has never been implemented at all. It wouldn't be particularly difficult to do, someone just has to mimic it from the sdl driver.

@sofistisoftware

1 is horribly wrong. This env var is supposed to be the user agent string, 1 is not a proper user agent AT ALL. Anyway, this env var shouldn't be used, because it overrides everything else (settings and per url settings, ...).

Now, about the actual issue, my guess if the user agent is not generated correctly on OS4 (the MorphOS version & co is generated dynamically, and if not done properly, it might give a string that some servers don't like, like google).

It doesn't happen on MorphOS, anyway. And if it did, it would simply mean google devs are stupid and do stupid user agent checks.

@jabirulo

There's no need to quit or anything after changing URL settings. They're applied in realtime.

Regarding the original issue about google using mobile version, unless OWB_USER_AGENT env var was overridden with something weird, it really shouldn't happen. If it does, you might want to delete cookies if they were set to something weird.

@kas1e

The MPlayer sources i provided have a proper fallback when overlay is not available (in that case, cgx_wpa is used instead).

I assume it doesn't work in the OS4 port because you reused that old p96pip mplayer driver which is buggy in that regard. Like in the cgx_overlay(_gui) driver, it should have a proper preinit function that checks for overlay ability. If it returns false when overlay is not available it gives a chance to mplayer to try another video output driver in the list instead of just failing...

This stalling connection issue is most likely due to the version kas1e used for his build (7,34,0). There's no such issue with 7,28,1, and it's likely a regression in curl itself.

@kas1e & elwood

In Odyssey (like in IBrowse), there are 3 different levels: global (in odyssey settings), per-url settings (in odyssey url settings) and temporary/per-tab (in odyssey menu).

The menu has priority over url settings, which itself has priority over global settings.

So of course, the menu is not supposed to show url or global spoof agent at all. It's not related.

@Raziel

All I can say is that this works in OWB MorphOS 1.23:

/**/
address OWB.1
OPEN NAME "http://www.google.com"

or if you want a new page

/**/
address OWB.1
OPEN NAME "http://www.google.com" NEWPAGE

@Raziel

I don't know how these OS4 URL preferences works, considering i don't see any of the required parameters for Odyssey, but anyway, Odyssey has proper OpenURL support, and you can see the related configuration in OWB.readme (except the OWB port name has been replaced by Odyssey in the OS4 port, to support the deprecated Reaction OWB port).

I quote from the readme:

- OpenURL configuration
Rexx Port: OWB
Show:SHOW
Screen To Front: SCREENTOFRONT
Open: OPEN NAME="%u"
Open in new window: OPEN NAME="%u" NEWPAGE

So if something doesn't work, it's down to these URL preferences.

@eroom

Maybe you didn't enable localstorage or cookie saving in Odyssey's settings.

Anyway, you could also just use youtube.js (most recent version, of course), there's no configuration needed, and it's less heavy than YoutubeCentre.

@kas1e

Nope, overlay as default doesn't make sense at all, because it uses the whole view space to be displayed, and you really DO want to see the original page + embedded video by default.

Besides, there are also cases where CSS transforms are applied to the video. Since the overlay mode is just about displaying the raw video, it ignores any postprocessing done to the video.

Or again, there can be multiple videos on a given page. It really can't be default, no.

Another way would be to render the video directly to the given target render rect (be it with overlay or not, it doesn't matter actually), bypassing cairo (which is what i do which the Flash player), but again, it's not perfect either, because it would prevent any postprocessing done to the video too (think about things as simple as subtitles or even rescaling for instance).

There's no ideal way to deal with that, except having a faster cairo.

@kas1e

autoplay means that the video is started automatically, not repeated.

If you want repeat option it's available in builtin HTML5 player anyway, just check it in the context menu.

Quote:

This 35% idle state is only on main page, not the other ones (videos ones).

Just to be clear, the two players are using my player anyway, it's just that HTML5 offers a default player, that can be customized with js/css like Youtube does, and it can make it slower with scaling/js overhead.

Also, i'm not sure what you mean with half the videos work with the builtin HTML5 mode. They all work, when Flash replacement has to be done. It may be more than 50% of the time (there are still many videos not available as HTML5 yet), and if HTML5 cookie is disabled, it would be 100%, even.

And finally, i insist on the importance of the overlay mode : on a pegasos2, with overlay enabled, playing a 360p video doesn't eat more than 35% cpu (maybe slightly more when buffering from network quickly, but that's it).