SDL Now Disables Mir By Default In Favor Of Wayland Compatibility

[phoronix]

Phoronix: SDL Now Disables Mir By Default In Favor Of Wayland Compatibility

With Mir focusing on Wayland compatibility now, toolkits and other software making direct use of Mir's APIs can begin making use of any existing Wayland back-end instead. GTK4 drops the Mir back-end since the same can be achieved with the Wayland compatibility and now SDL is now making a similar move...

http://www.phoronix.com/scan.php?page=news_item&px=SDL-Mir-No-Longer-Default

[paupav]

It would be great to see difference in fps performance between Mir/Mir and Mir/Wayland. I remember having very high fps animations on Unity8.
Everything was so buttery. Ahh, good times.

[Kemosabe]

It would be great to see difference in fps performance between Mir/Mir and Mir/Wayland. I remember having very high fps animations on Unity8.
Everything was so buttery. Ahh, good times.

Yeah sadly the numbers spoke for Mir ...

[ElectricPrism]

Unification is good, Mir Compositor using Wayland protocol will do just fine.

[Guy1524]

Why though? What is the issue with having the Mir support.

[ssokolow]

Why though? What is the issue with having the Mir support.

More code for the SDL people to maintain and more regression testing to do before each release.

(I've always felt that, if Canonical wanted everyone to accept patches for their NIH Wayland competitor rather than improving the Wayland offerings, just because they wanted an alternative to Wayland that they could sell under an alternative license agreement, then it should be contingent on Canonical paying or donating time and resources to make up for the additional maintenance and testing burden.)

[paupav]

Yeah sadly the numbers spoke for Mir ...

Because canonical developers are good or because original design is still present and it needs just some refactoring or it needs refactoring ? hmm

[Delgarde]

Why though? What is the issue with having the Mir support.

It's the same issue that applies to all code – someone needs to maintain it. SDL isn't static and unchanging, and so when they change things, they need to ensure that nothing breaks. And that's easy with the core code that always gets run and tested, but not easy when it's code that exists only to support some obscure platform that the core developers don't use and don't care about.

This was always one of the problems with Mir – even if Canonical were going to do the work to add Mir support to various toolkits, it's not enough to just to a one-time code drop to add Mir support. Supporting a new platform requires a long-term commitment...

[varikonniemi]

Yeah sadly the numbers spoke for Mir ...

If someone can demonstrate this after wayland support is mature, i think we have a fundamental problem with wayland's design. But i don't think anyone has demonstrated such...

[cybertraveler]

If someone can demonstrate this after wayland support is mature, i think we have a fundamental problem with wayland's design. But i don't think anyone has demonstrated such...

"design" is the keyword there. Any performance comparisons between Mir and Wayland would have actually been a comparison between the 1 and only Mir implementation and 1 of many Wayland implementations. Wayland is a protocol. It would have been implementations of Wayland that were benchmarked against Mir. If Wayland performed poorly it would tell you nothing about which design was superior.