[nitrogen]

It may sound reasonable, but ultimately is against the universal spirit of the web, and thus should not be standardized. Your #1 should be No, because if it were Yes, we'd have a mess of mutually incompatible, vendor-specific, proprietary languages fragmenting 3D content on the web.

[snprbob86]

Mutually incompatible, vendor-specific, proprietary anything is an inevitability. #1 isn't about that. That's what #2 and #3 are about.

#1 is about planning for extensibility. Just look at the hackery with JS where lonely, otherwise ignored, strings are used for things like "use strict" and "use asm". Or where Microsoft added "conditional comments", which quite frankly, was essential to the development of Outlook Web Access, which basically gave us Ajax. Or all the absurd vendor prefixes on CSS tag names. Or one of 100 other little hacks that browser vendors have invented to try to innovate past the standard. Pushing pass the standard, by the way, is the only way forward. We've learned that lesson by now, so we should plan for extensibility.

[nitrogen]

OpenGL already has a mechanism for extensibility, and proprietary junk is only an inevitability if we allow it to be enshrined in open standards. There is no reason to accept proprietary DRM plugins in CDM, and there is no reason to accept proprietary shader languages.

The reasons are manifold, but here are a few:

- Standardizing non-standardness gives proprietary implementations an unwarranted air of legitimacy and blesses incompatibility.

- Proprietary plugins and extensions are more likely to have untested security vulnerabilities and widen the browser attack surface.

- Proprietary extensions violate the essential web principles of cross-platform compatibility, graceful degradation, progressive enhancement, and accessibility.