Matroska - Learn it, know it

[Atamido]

The public beta release of Matroska is inching closer and closer every day now. With this impending release, we would like to have a kind of FAQ for the general populace to be able to glance at to answer basic questions, without needing to go into to much technical detail. Now, the doom9 crowd is obviously above average for anything video related, and so its a good place to start for questions.

I would encourage everyone to look at the Matroska homepage (http://www.matroska.org) to get a good idea of the purpose of Matroska. There is also an XML representation of the structure of a Matroska file with all of the fields that have been defined here. (http://www.matroska.org/Matroska9.XML) Information on what each 'element' is for can be found in the specs page. (http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/*checkout*/matroska/doc/website/specs/index.html?rev=HEAD&content-type=text/html)

Please read as much as you feel comfortable, and post questions that you have about Matroska. Nothing is to small, or to great, and everything posted will help tremendously.

[alexnoe]

Mh... I wonder about the language!
It specifies a binary and octet aligned format What do you mean with "octet"?
8 Bytes are called a "quadword", or simply "_int64", but not an octet. "Un octet" is one byte. However, if you refer to bytes (as in "float (sizes of 4, 8 and 10 octets)"), then you should say "byte" and not "octet"...unless you want to mix english and french.
informationsWhen did "information" get a plural form?

[Atamido]

Originally posted by alexnoe
What do you mean with "octet"?
8 Bytes are called a "quadword", or simply "_int64", but not an octet. "Un octet" is one byte. However, if you refer to bytes (as in "float (sizes of 4, 8 and 10 octets)"),
Oddly, this did come up earlier, and I guess it does bare clarification in the specs.

The term 'octet' refers to an 8 bit, byte. The reason for the use of this term is because a byte can be of varying number of bits, depending on the system that is being used.
then you should say "byte" and not "octet"...unless you want to mix english and french.
For some reason, all of the german and french members of the Matroska team thought the same thing. However, the word 'octet' is an english word. Everyone seems to understand what it means, but then point at the french. (I'm in/from the US)

When did "information" get a plural form? Must be a french spelling? ;) Will get this taken care of asap.


Keep those comments coming!

[robUx4]

Yeah, all documents must be full of incorrect english :)

But on the advises of Frank Klemm (MPC coder) we decided to chose octet which is the only word that refers to 8 bits for sure. And it's a word in english, french and german and probably other languages too. That's the only word we use in french to refer to 8 bits. (I'm french)

[alexnoe]

OK, so "octet" is another word which has been copied from french...so you use it to express that you do not make the same crap as someone did in the DTS format ;)
In German, it would be Oktett, but using it to refer to 8 bits would be weird.

OK, now reading more closely:
This document is the not the real format specification. Mh....
But since it's rather completeIs that supposed to mean "quite complete"? => German mistake (unless it is negative that it is almost ready...then "rather" would of course be fine)
But a frame with a past timecode must reference a frame already know, otherwise it's considered as bad/void.considered as? I thought "to consider" would be used without preposition :confused:
Lacing is mechanism to save space when storing data. Isn't there missing something?

[ChristianHJW]

LOL !!! LMAO !!! ROTFL !!!

Hey alexnoe, thats big fun bashing Steve for his crappy english :D :D :D .. time for me to read the specs and help you a bit with that .... :cool:

ChristianHJW feels robux4' whip circling over his head and ass ...

[Atamido]

Okay, I went ahead and performed a spell check on the specs, so now you should be able read through without getting bogged down by the gramatical problems.

And BTW, the word 'informations' occurred no less than 10 times. Those darn french. :p


Keep it coming!



Edit: Specs updated

[alexnoe]

@Pamel: You can't blame the French people for their language...not long ago, someone on cdfreaks wrote "it's the better way" (he wanted to express "c'est la meilleure solution")
Lacing sizes: only the first one will be coded, 800 gives 255;255;255;35. The size of the last frame is deduced from the total size of the Block.And I thought that German would be the only language where all nouns are capitalized :confused:
Multi: specify that the element can be found multiple times in the upper level elementI would actually consider "multi" a singular subject...
Bit 6 and 5 of the Block gives the following behavior :Mh...

[Atamido]

1. I always blame the french. They are an easy target, and I don't have to worry about retalition because the haven't won a major war since "that short dude" was in power.

2. In this case "Block" is a name, or proper noun, and so it is capitalized.

3. Corrected.

4. I corrected that, and then realized its not relevant because the specs for it changed. Removed.

Great comments. Keep those coming.

Does anyone have any questions not related to spelling and gramar?
Edit: Spelling...

[bond]

Originally posted by Pamel
I always blame the french. They are an easy target, and I don't have to worry about retalition because the haven't won a major war since "that short dude" was in power.So a small country like france could rule half of the world without winning a "major war" ;)

Does anyone have any questions not related to spelling and gramar?yes when will there be the first public beta release of matroska :D :rolleyes:

[robUx4]

Originally posted by bond
So a small country like france could rule half of the world without winning a "major war" ;)

Talking about Napoleon ? Or Lafayette ?

Anyway, back on topic. There are many places in the specs where some words contain upper cases. It's because the word refer to a matroska element, not just the word. So it's not a typo.

[Atamido]

Originally posted by robUx4
Talking about Napoleon ? Or Lafayette ?

I was referring to Napoleon, taking a cue from "Bill and Ted's Excellent Adventure."

[bond]

Originally posted by robUx4
Talking about Napoleon ? Or Lafayette ?nope i talked about france and its colonies :p

but enough of history...

[BlackSun]

IMHO octet is less confusing than byte/bit

[juicemansam]

Orignally posted by Pamel
The term 'octet' refers to an 8 bit, byte. The reason for the use of this term is because a byte can be of varying number of bits, depending on the system that is being used.

From what I remember from my CS classes, a BYTE is defined as 8 BITS. I've never heard of a BYTE being able to be a varying number of BITS, until now. Then again, I haven't taken any classes involving architectures, other than ix86's, where this might be true.

Maybe since I've taken several programming language classes, including Assembly, I only see the down-n-dirty definitions. That leads me to believe that a BYTE is not a Data Type, but a Data Size. I know that higher-level languages like C or C++ will have varying sizes for the Data Types with the same name when refering to different architectures.

Well that's my 2 cents.

[alexnoe]

The size of a "word" depends on the machine; the size of a byte does not

[Atamido]

Errr.... the size of a byte isn't necessarily 8 bits, though that is the norm now. Check here (http://wombat.doc.ic.ac.uk/foldoc/foldoc.cgi?query=byte) for more information.

A little note has been added to the specs that clarifies what an octet is.

[robUx4]

At least people will learn one thing by reading the matroska specs :)
They should be encouraged to do so ! ;)