Comment 70 for bug 346289

Hello Gentlemen,

To hopefully quell the amount of comments this bug report is getting, I've taken some time to fill in the description of this bug report with a summary of the situation. Please take a moment to read it.

Really, this is not a video driver bug specifically, but rather an issue with how Adobe Flash is implemented to *use* the video driver. That's why the behavior can vary depending on the type or version of video driver being tested. Also, really it's not just one specific bug, but a collection of issues several of which are fundamental design flaws in Adobe's implementation. Adobe has provided some nicely written and illustrated materials explaining these design issues, which I've linked to in the description and would encourage you to check out.

I am not sure there is really anything we can do at the distro level. There are a couple workarounds I can think of, but I have a feeling they'd end up causing more breakages than they solve. It would be interesting to experiment with them, but unfortunately my plate is quite full with other projects so I will leave the experiments to any of you that would like to indulge your curiosity.

In the one case, where we would force OverrideGPUValidation=true as a default in flashplugin-nonfree, this is quite straightforward to implement - just include a config file in the flashplugin-nonfree debian/ directory to be installed to /etc/adobe/ when the package is installed. The challenge is that we would need this tested across a breadth of different video driver combinations, with and without Compiz and/or Unity enabled. This could be crowd-sourced, with results tabulated in a wiki similar to how we've done with other testing (see the results table on https://wiki.ubuntu.com/X/Testing/NouveauEvaluation ... something akin to that would be nice).

The other option I can think of would be to redirect the 'fullscreen' action in Flash, to launch Totem or whatever the default video player is. For this option, we'd need to see a working proof of concept that folks could play with. This has some obvious tradeoffs in terms of usability so the benefit would need to be quite striking. I think this would be a fairly straightforward project one of you could do.

There could well be other options as well I've not thought of. In any case, what we need are proof of concepts that demonstrate that they do indeed work. Please keep in mind that what works for you may not work as well for someone else, but for Ubuntu we need to address as broad a range of user needs as possible, so the more testing you can do of your concept, the better.

I hope all this information is useful, sorry this is not an issue we can trivially fix at the distro level. I am hopeful that this explanation inspires someone to produce a good proof of concept of some better solution, that would let the Desktop team invest resources into getting it integrated into Ubuntu proper for everyone.