4 ms·

Given that there is no sensible way to implement DRM in open source, the W3C should not approve the standard. Or the other way around, W3C should only approve

by phicoh 10y ago

Given that there is no sensible way to implement DRM in open source, the W3C should not approve the standard. Or the other way around, W3C should only approve feature complete standards that can be implemented using open source. You cannot have an open web if an essential part is governed by binary blobs. Nothing is lost if W3C does not approve an interface to secret technology.

gcp 10y ago
EME can be implemented in open source (there's an implementation in Chromium and another one in Firefox). The problem is the CDM, which the W3C isn't standardizing.
- JoshTriplett 10y ago
  Which means you can't actually ship any useful implementation, since it won't have a backend to talk to. And an Open Source implementation would inherently be non-functional.
  - gcp 10y ago
    You can implement EME in open source (the part that W3C standardized), and then throw a closed source CDM module on top of it. You don't even have to provide the CDM, third parties can provide multiple ones. This is in fact how it works in Firefox. There's no part of Firefox that is closed source, including EME support or even the CDM sandbox. The CDM is loaded at runtime from an external server. Focusing on the W3C and/or it's relation with open source is completely and utterly missing the point, but it's explained enough throughout this thread that I see no point in repeating it once again here.
    - JoshTriplett 10y ago
      > You can implement EME in open source (the part that W3C standardized), and then throw a closed source CDM module on top of it. Which means the whole thing doesn't work in a fully Open Source browser. EME is non-functional without a proprietary CDM. It's like standardizing the <object> or <applet> tags: yes, they're a standard, but they're a standard way to talk to completely non-standard bits. But worse, because you could use <object> or <applet> to talk to an Open Source plugin, but a CDM completely loses what little function it has if open. (That includes hardware-backed CDMs, since signed software you can't replace isn't Open Source either.)
- phicoh 10y ago
  That's why I wrote 'feature complete standards'. Anybody can come up with half finished standards that are only there to serve some proprietary technology. An interface to an unspecified DRM module is not a useful web standard. And due to the nature of DRM, a fully specified DRM standard is also completely useless. DRM is just not compatible with open.

by phicoh 10y ago

gcp 10y ago

EME can be implemented in open source (there's an implementation in Chromium and another one in Firefox). The problem is the CDM, which the W3C isn't standardizing.

JoshTriplett 10y ago
Which means you can't actually ship any useful implementation, since it won't have a backend to talk to. And an Open Source implementation would inherently be non-functional.
- gcp 10y ago
  You can implement EME in open source (the part that W3C standardized), and then throw a closed source CDM module on top of it. You don't even have to provide the CDM, third parties can provide multiple ones. This is in fact how it works in Firefox. There's no part of Firefox that is closed source, including EME support or even the CDM sandbox. The CDM is loaded at runtime from an external server. Focusing on the W3C and/or it's relation with open source is completely and utterly missing the point, but it's explained enough throughout this thread that I see no point in repeating it once again here.
  - JoshTriplett 10y ago
    > You can implement EME in open source (the part that W3C standardized), and then throw a closed source CDM module on top of it. Which means the whole thing doesn't work in a fully Open Source browser. EME is non-functional without a proprietary CDM. It's like standardizing the <object> or <applet> tags: yes, they're a standard, but they're a standard way to talk to completely non-standard bits. But worse, because you could use <object> or <applet> to talk to an Open Source plugin, but a CDM completely loses what little function it has if open. (That includes hardware-backed CDMs, since signed software you can't replace isn't Open Source either.)
phicoh 10y ago
That's why I wrote 'feature complete standards'. Anybody can come up with half finished standards that are only there to serve some proprietary technology. An interface to an unspecified DRM module is not a useful web standard. And due to the nature of DRM, a fully specified DRM standard is also completely useless. DRM is just not compatible with open.