NSS Updated On Ubuntu 12.04/14.04 To Allow Netflix Support

[AJenbo]

Right, I agree. If such information were present in the UA string, then of course that would be valuable.
We are, however, talking about the Chrome UA string on Linux, which does not contains such detailed version information.

You fail to see it from Netflix point of view. The UA does contain the relevant information 98% of the time. There is no other way of getting this information. Thats why it was set up, and that is why it has to be adjusted to account for Linux now generally working, rather then generally not working.

[anda_skoa]

You fail to see it from Netflix point of view. The UA does contain the relevant information 98% of the time.

And it is not used in this 98% of the time, but used for the 2% where it is not applicable?

How does that make sense from any point of view?

As I said it makes sense to use the UA when it contains information that allows to do pre-filtering, but using it when it does not contain any relevant information is not ony useless, it is stupid.

There is no other way of getting this information.

Yes, there is.
The whole point of the EME specification is to make it possible for a web page code to query the CDM and initiate key exchange.

Which the web site would do if it where interested.

Thats why it was set up, and that is why it has to be adjusted to account for Linux now generally working, rather then generally not working.

If that were the case (which it isn't), why was it used to make it generally not working?

I really don't get why this is so hard to understand.
If one wants to know if a certain vehicle is operational, one can check the license plate first and conclude if they are missing that it is not.
But once that simple pre-check is passed, the only plausible next steps are checking if it has an engine and ultimately trying to start it.
Doesn't mean you have to start driving right away, right?

The useless UA check we are discussing here is like seeing the license plate is from a certain country and deciding that it can't probably be working, inspite the very same model from any other country is known to work.

Cheers,
_

[AJenbo]

A car metaphor, really? I'm not even going to address that.

And it is not used in this 98% of the time, but used for the 2% where it is not applicable?

How does that make sense from any point of view?

As I said it makes sense to use the UA when it contains information that allows to do pre-filtering, but using it when it does not contain any relevant information is not ony useless, it is stupid.


Yes, there is.
The whole point of the EME specification is to make it possible for a web page code to query the CDM and initiate key exchange.

If the check gives a false positive depending on the underlying operating system it leaves them with no alternative.

If that were the case (which it isn't), why was it used to make it generally not working?

To be on the safe side (not necessarily the side you would think is the best dessication). It's a pragmatic solution, to a problem that there may or may not be a better solution for.

[anda_skoa]

A car metaphor, really? I'm not even going to address that.

Yes, excuse me, but since people seem to have a really hard time to understand the basics involved, breaking to down to a non-IT metaphor looked like a viable option.

I understand that it is hard to comprehend that computers do not perform magic and software cannot make decisions based on non-existant information anymore than a human could.
A data set that does not contain the information a program would need for decision making is not a viable input for that program.
Like in this the User Agent String. It does not contain platform version information regarding the decision at hand, so it can't be used.

If the check gives a false positive depending on the underlying operating system it leaves them with no alternative.

It can neither give a false nor a true positive, nor a false or true negative, it simply does not contain enough information.

It's a pragmatic solution, to a problem that there may or may not be a better solution for.

If it is a solution, it is certainly not for the problem of determining whether the browser is a viable one.

It is a solution to the "problem" of not wanting customers.

However, it is being spin-doctored to be a solution for the other problem.
It is of course everyone's personal choice to believe in things refuted by easily available facts.

Cheers,
_

[AJenbo]

A data set that does not contain the information a program would need for decision making is not a viable input for that program.
Like in this the User Agent String. It does not contain platform version information regarding the decision at hand, so it can't be used.

You seam to be of the misconception that the filter is for handling Linux users. It's not, it was put there to filter for IE11 on Win8, Safari on OS X 10.9 and ChromeOS, this information IS in the UA. It's a whitelist and Linux is simply not on it. They probably didn't even stop to consider Linux when making the list, but now it has to be adjusted to allow Chrome on Linux to work out of the box, and they are willing to do so if the majority of Linux users have the appropriate software available to them.

[anda_skoa]

You seam to be of the misconception that the filter is for handling Linux users. It's not, it was put there to filter for IE11 on Win8, Safari on OS X 10.9 and ChromeOS, this information IS in the UA. It's a whitelist and Linux is simply not on it.

Exactly, thank you!

Hence there not being any relation what so ever between some distribution updating some libs and this check.

Cheers,
_

[AJenbo]

Exactly, thank you!

I'm glad you understand now.

Hence there not being any relation what so ever between some distribution updating some libs and this check.

There is one, Netflix isn't going to bother changing it before it works for the majority of the affected users. I think this is much more an issue of management then marketing.

[anda_skoa]

I'm glad you understand now.

Ah, you are misunderstanding me.
I obviously knew that, but many other had the impression that it was somehow related to Chrome's CDM not working on all possible systems.

The UA check always was about Linux in general, never about whether it would work.

There is one, Netflix isn't going to bother changing it before it works for the majority of the affected users.

The majority of Linux users will remain Linux users, I doubt any thing so easily trickable as a UA check will change that.

I think this is much more an issue of management then marketing.

I was referring to the trick of spinning the "no linux" check as somehow related to two specific versions of one distribution being broken regarding Chrome's requirements for its CDM to work properly.

Given a position where you are explicitly shutting out Linux users despite their systems' capabilities and getting people not only to believe it was somebody else fault but making them distribute that spin is exceptionally good marketing. Wouldn't you agree?

Cheers,
_