From nobody Tue Dec 8 17:43:32 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 5D4DC1A6FDF; Tue, 8 Dec 2015 17:43:30 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.91 X-Spam-Level: X-Spam-Status: No, score=-1.91 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, T_RP_MATCHES_RCVD=-0.01] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id yJmJx47sT4YO; Tue, 8 Dec 2015 17:43:28 -0800 (PST) Received: from nostrum.com (raven-v6.nostrum.com [IPv6:2001:470:d:1130::1]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id E07D01A6FC8; Tue, 8 Dec 2015 17:43:21 -0800 (PST) Received: from [10.0.1.10] (cpe-70-119-203-4.tx.res.rr.com [70.119.203.4]) (authenticated bits=0) by nostrum.com (8.15.2/8.14.9) with ESMTPSA id tB91hKQm086672 (version=TLSv1 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO); Tue, 8 Dec 2015 19:43:21 -0600 (CST) (envelope-from ben@nostrum.com) X-Authentication-Warning: raven.nostrum.com: Host cpe-70-119-203-4.tx.res.rr.com [70.119.203.4] claimed to be [10.0.1.10] From: "Ben Campbell" To: draft-ietf-codec-oggopus.all@ietf.org, codec@ietf.org Date: Tue, 08 Dec 2015 19:43:21 -0600 Message-ID: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit X-Mailer: MailMate (1.9.3r5187) Archived-At: Subject: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 09 Dec 2015 01:43:30 -0000 (oops, sent from wrong address, and got bounced from a lot of lists. Apologies for the repeat. Please send replies to this address.) Hi, Here is my AD evaluation of draft-ietf-codec-oggopus-09. Codec encapsulation is not my strong suit, so I suspect many of my comments can be answered with "this is the way we normally do this thing". But I'd still like to discuss these before going to IETF last call. Thanks! Ben. ---------------- Substantive: ============ - section 2, last paragraph: Is this paragraph necessary or useful? (And is it really impossible to create a non-compliant implementation that meets all the MUSTs?) I suggest removing it. - section 3: This section seems to use 2119 keywords to describe what I _think_ are existing Ogg requirements. If so, they should use descriptive text, unless they are in the form of attributed quotes. - 3, last paragraph: -- "implementations need to be prepared": maybe that should be a MUST or SHOULD? -- "last page SHOULD NOT be a continued packet": Why not MUST NOT? Can you describe reasons someone might reasonably violate this? - 4.2, 2nd paragraph: -- "samples which SHOULD be skipped": Why not MUST? (also, s/which/that) -- "SHOULD NOT be played": Is this redundant with SHOULD be skipped? - 4.3: Should the reference to [vorbis-trim] be normative? Can the feature be implemented without understanding the referenced doc? - 4.6: Should the reference to [seeking] be normative? Can the section be implemented without understanding it? - 5.1, list entry 5: What is the unit for "Input Sample Rate"? - 5.1, entry 6, sentence starting with "Virtually all players..." Is this a normative statement, or a statement of fact? If the former, consider dropping “Virtually all” - 5.1, entry 6, last paragraph: -- "which MUST be omitted when the channel mapping family is 0": That seems redundant to the previous paragraph. -- "Implementations SHOULD reject ID headers" What does it mean, in context, to "reject" headers? - 5.1.1.4: Why not MUST? - 5.2, last two paragraph: Why are the two SHOULDs not MUSTs? - 5.2.1, 2nd to last paragraph, last sentence: I don’t think muxers assume things—what is the required concrete behavior? - 5.2.1, last paragraph: Why is the SHOULD NOT not MUST NOT? -6, first paragraph: Why are the two SHOULDs not MUSTs? What does it mean to reject a packet? -7: "reference implementation" could use a citation. Should [linear-predection] be normative? Can this be implemented without understanding that? If it does need to be normative, can you find a more stable reference? - 9, first paragraph: This seems to defer at least some security considerations to 4732. If so, this should be a normative reference. If, on the other hand, 4732 just provides additional information that is not necessary to understand this section, then the wording should be adjusted. - 13: Is this needed? Are the IETF BCP 78 and 79 provisions incorrect for this? Editorial ========= - Section 1: Consider spelling out multiplexer (muxer) and demultiplexer (demuxer) on first mention. - 3: Please expand TOC, SILK, and CELT on first mention. - 5, steps 2 and 3: These steps in combination seem to say “ decode at the highest available supported rate that the hardware can handle.” If so, the wording could be simplified. - 7, 2nd paragraph: I don’t think that’s literally true, you’ve defined 0, 1, and 255. The rest are reserved. (I guess one could argue the guidance to treat reserved values as 255 supports this statement, but I think that’s a stretch. -5.1.1.2: Do you expect the reader to know what Vorbis channel order is? Also, please expand FLAC on first mention. - 5.2.1, paragraph starting with " An Ogg Opus stream MUST NOT have more than one of each tag, and if present their values MUST be an integer" each of these two new tags, or all tags? (I assume the former from the numeric values) Also, singular/plural mismatch in "their values MUST be an integer" -10, last paragraph: Updates, or extends? If the former, the draft needs an “Updates RFC 5334” field in the initial heading. From nobody Fri Dec 11 14:16:52 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C60541A9068; Fri, 11 Dec 2015 14:16:50 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.613 X-Spam-Level: X-Spam-Status: No, score=-2.613 tagged_above=-999 required=5 tests=[BAYES_50=0.8, HELO_MISMATCH_ORG=0.611, HOST_MISMATCH_COM=0.311, RCVD_IN_DNSWL_HI=-5, SPF_SOFTFAIL=0.665] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7ZPJryj_zlgU; Fri, 11 Dec 2015 14:16:43 -0800 (PST) Received: from smtp.mozilla.org (mx1.scl3.mozilla.com [63.245.214.155]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id A76FE1A908A; Fri, 11 Dec 2015 14:16:43 -0800 (PST) Received: from localhost (localhost6.localdomain [127.0.0.1]) by mx1.mail.scl3.mozilla.com (Postfix) with ESMTP id 34751C1852; Fri, 11 Dec 2015 22:16:43 +0000 (UTC) X-Virus-Scanned: amavisd-new at mozilla.org Received: from smtp.mozilla.org ([127.0.0.1]) by localhost (mx1.mail.scl3.mozilla.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id hUXwMJwIPICA; Fri, 11 Dec 2015 22:16:41 +0000 (UTC) Received: from [10.130.20.84] (unknown [207.190.10.41]) (Authenticated sender: tterriberry@mozilla.com) by mx1.mail.scl3.mozilla.com (Postfix) with ESMTPSA id AF02EC1812; Fri, 11 Dec 2015 22:16:40 +0000 (UTC) Message-ID: <566B4B47.9010809@xiph.org> Date: Fri, 11 Dec 2015 14:16:39 -0800 From: "Timothy B. Terriberry" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:29.0) Gecko/20100101 SeaMonkey/2.26 MIME-Version: 1.0 To: Ben Campbell , draft-ietf-codec-oggopus.all@ietf.org, codec@ietf.org References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> In-Reply-To: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> Content-Type: multipart/mixed; boundary="------------090501040108020500070205" Archived-At: Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 11 Dec 2015 22:16:50 -0000 This is a multi-part message in MIME format. --------------090501040108020500070205 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Ben Campbell wrote: > Here is my AD evaluation of draft-ietf-codec-oggopus-09. Codec > encapsulation is not my strong suit, so I suspect many of my comments > can be answered with "this is the way we normally do this thing". But > I'd still like to discuss these before going to IETF last call. Thank you for the very thorough review. Comments in-line below (as an individual). > - section 2, last paragraph: > > Is this paragraph necessary or useful? (And is it really impossible to > create a non-compliant implementation that meets all the MUSTs?) I > suggest removing it. This was copied from RFC 5334. I don't think anyone has any particular attachment to it, so let's just remove it. > - section 3: > This section seems to use 2119 keywords to describe what I _think_ are > existing Ogg requirements. If so, they should use descriptive text, > unless they are in the form of attributed quotes. I think I can attribute this mostly to paranoia over RFC 3533 being informational, and using descriptive text of its own for what should be normative requirements. E.g., "Ogg also allows but does not require secondary header packets after the bos page for logical bitstreams and these must also precede any data packets in any logical bitstream. These subsequent header packets are framed into an integral number of pages, which will not contain any data packets." (from RFC333) vs. "The second packet in the logical Ogg bitstream MUST contain the comment header ... It MAY span multiple pages, beginning on the second page of the logical stream. However many pages it spans, the comment header packet MUST finish the page on which it completes." (from this draft) Unless someone objects, though, I'll try to rewrite the latter without the 2119 keywords. > - 3, last paragraph: > -- "implementations need to be prepared": > maybe that should be a MUST or SHOULD? I think the intent was just to point out that someone might violate the preceding SHOULD. It's not clear what actionable thing we'd be saying they MUST do (it would depend on the particular implementation). > -- "last page SHOULD NOT be a continued packet": > Why not MUST NOT? Can you describe reasons someone might reasonably > violate this? The intent was to allow but discourage page-level editing of files (i.e., cutting and splicing files by copying complete pages without examining the packets on them). Even just saving a live stream that drops at some point, you might reasonably expect to just be able to copy the data you received to a file. A MUST here would mean you would have to parse the pages, identify the continued packets from the lacing values, remove the extraneous data from the last page if you found some, repack the page buffer, and recompute the page checksum. That's a lot more machinery, and people would likely just ignore the requirement in practice. See also the discussion at . > - 4.2, 2nd paragraph: > -- "samples which SHOULD be skipped": Why not MUST? (also, s/which/that) Some specific applications might have a reason for looking at that data (for example, when you have multiple files that you know were produced by chopping up a single existing file without re-encoding). In practice we haven't needed anything stronger to convince people to implement this correctly (especially when we point them to ). > -- "SHOULD NOT be played": Is this redundant with SHOULD be skipped? Yes. Deleted. > - 4.3: Should the reference to [vorbis-trim] be normative? Can the > feature be implemented without understanding the referenced doc? The answer to the second question is "yes", which I think makes the answer to the first question "no". The reason this is here is to tell people familiar with how Vorbis works what mechanism they should use to achieve the same outcome in Opus. But if you're not familiar with Vorbis, then you can safely ignore the whole paragraph. > - 4.6: Should the reference to [seeking] be normative? Can the section > be implemented without understanding it? It is possible to implement seeking without knowing anything other than what is described in RFC 3533, but the linked documentation does help considerably. I would argue that the section itself is not normative, in that you could implement seeking with some other method (for example, linearly scanning the whole file on open to build an index, possibly caching such indexes for files that are opened frequently: this would not be my recommended approach, but people have done it in practice). You could also choose to implement only approximate seeking by just guessing a file position and decoding from there (also done in practice by some players, and a common method for other formats like MP3). Or you could choose not to implement seeking at all. > - 5.1, list entry 5: What is the unit for "Input Sample Rate"? Hz. It was specified in Figure 1, but you're right, it should be in the text, too. Added the sentence, "This is the sample rate of the original input (before encoding), in Hz." > - 5.1, entry 6, sentence starting with "Virtually all players..." > Is this a normative statement, or a statement of fact? If the former, > consider dropping “Virtually all” Dropped "virtually all". > - 5.1, entry 6, last paragraph: > -- "which MUST be omitted when the channel mapping family is 0": That > seems redundant to the previous paragraph. Yes. Deleted the sentence in the previous paragraph. > -- "Implementations SHOULD reject ID headers" > What does it mean, in context, to "reject" headers? The same thing as rejecting a stream (i.e., don't try to play/decode it). Changed to say that instead. > - 5.1.1.4: Why not MUST? I think the idea was to allow assigning reserved values without explicitly updating this specification. Even though I expect we _would_ want to update this specification, given how long it's taken to get this draft published, I'm wary of adding more pressure to people not to deploy new mappings (because this RFC says they MUST not), even when it looks like they'll be relatively stable and interoperable. At least not for the sole reason that they haven't gotten an RFC number yet. > - 5.2, last two paragraph: Why are the two SHOULDs not MUSTs? For the first SHOULD, this is the same as above: some other specification may eventually define the format of the data here and how to modify it, and I didn't think we'd want to make people supporting that specification in violation of this specification. I'm also leaning towards keeping this one a SHOULD since a) the eventual format of this data may apply to all codecs which share the vorbiscomment style of metadata, and b) we would probably try to do this informally first to gain implementation experience, and only bring it to the IETF if there is sufficient interest. I didn't want to make updating this document a prerequisite for that. See also the discussion in this thread: (skip to for the resolution of the debate). For the second SHOULD, it felt like the right strength for a judgment call like "excessive". I have no objection to changing it to MUST. > - 5.2.1, 2nd to last paragraph, last sentence: > I don’t think muxers assume things—what is the required concrete behavior? I replaced this with "A muxer SHOULD place the gain it wants other tools to use by default into the 'output gain' field, and not the comment tag." > - 5.2.1, last paragraph: Why is the SHOULD NOT not MUST NOT? In some contexts, there may be no confusion with multiple normalization schemes. The language in this section was debated a good deal on the list (see threads , , and ). I believe everyone is satisfied with the current compromise, and would prefer not to change the strength of any normative statements in it to avoid re-igniting the debate. > -6, first paragraph: > Why are the two SHOULDs not MUSTs? What does it mean to reject a packet? For the first SHOULD, the fact that most decoders are likely to reject such packets is enough to discourage encoders from emitting them, and I was wary of imposing a MUST restriction when RFC 6716 chose not to do so. For the second, this is the same as the previous comment about excessive allocations. I have no objection to changing this one, either. "Reject a packet" is just shorthand for "treat them as if it was a malformed Opus packet with an invalid TOC sequence". I've added explicit text saying the latter. > -7: "reference implementation" could use a citation. Should Added a reference to RFC 6716. > [linear-predection] be normative? Can this be implemented without > understanding that? If it does need to be normative, can you find a more > stable reference? Linear prediction is something that would be understood by "a person having ordinary skill in the art" (to borrow a phrase). We just wanted to include a reference to help those who might not be DSP engineers but wanted to learn more. > - 9, first paragraph: > This seems to defer at least some security considerations to 4732. If > so, this should be a normative reference. If, on the other hand, 4732 > just provides additional information that is not necessary to understand > this section, then the wording should be adjusted. Upgraded the reference to normative. > - 13: > Is this needed? Are the IETF BCP 78 and 79 provisions incorrect for this? This text was specifically requested by the Debian project (and I support it personally). The same issue came up for RFC 6716 (for the same reasons). See the discussion at and . Reviewing those conversations, I do notice that the language of that section was changed after last call for RFC 6716, but the corresponding language in our source repository was not updated (because it was added to the document boilerplate manually by the RFC Editor, and not included in a separate section). > Editorial > ========= > - Section 1: Consider spelling out multiplexer (muxer) and demultiplexer > (demuxer) on first mention. Done. > - 3: Please expand TOC, SILK, and CELT on first mention. Doing this directly in the current text was a little bit awkward. I reworded it to "...An implementation of this specification SHOULD treat any Opus packet whose duration is different from that of the first Opus packet in an Ogg packet as if it were a malformed Opus packet with an invalid Table Of Contents (TOC) sequence. "The TOC sequence at the beginning of each Opus packet indicates the coding mode, audio bandwidth, channel count, duration (frame size), and number of frames per packet, as described in Section 3.1 of [RFC6716]. The coding mode is one of SILK, Hybrid, or Constrained Energy Lapped Transform (CELT). The combination of coding mode, audio bandwidth, and frame size is referred to as the configuration of an Opus packet." I did not expand SILK because it is not an acronym. > - 5, steps 2 and 3: > These steps in combination seem to say “ decode at the highest available > supported rate that the hardware can handle.” If so, the wording could > be simplified. It is a little bit more complicated. There are two rates here: one used for decoding, and one used for playback. The difference between steps 2 and 3 is that in step 3, the hardware's highest available sample rate is not one of the rates supported by Opus decoding. For example, if the highest available sample rate the hardware can handle is 32 kHz, then this is not a "supported rate" (i.e., not supported by Opus decoding), so step 2 does not apply and step 3 does. Step 3 says to decode at the next highest "supported rate", which would be 48 kHz, and then resample to a rate the hardware can handle (in this case 32 kHz). I don't think that's equivalent to your summary of the two steps. > - 7, 2nd paragraph: > I don’t think that’s literally true, you’ve defined 0, 1, and 255. The > rest are reserved. (I guess one could argue the guidance to treat > reserved values as 255 supports this statement, but I think that’s a > stretch. Yes, this is really true for the currently defined mapping families. Changed to "Each currently specified value of this octet indicates..." > -5.1.1.2: Do you expect the reader to know what Vorbis channel order is? > Also, please expand FLAC on first mention. It's a fairly common term among those who work with audio formats, but I added a "(see below)" for those who don't, since the table below it lists what that order is. > - 5.2.1, paragraph starting with " An Ogg Opus stream MUST NOT have more > than one of each tag, and if present their values MUST be an integer" > > each of these two new tags, or all tags? (I assume the former from the > numeric values) Also, singular/plural mismatch in "their values MUST be > an integer" "each of these tags" seems to resolve both issues. > -10, last paragraph: > > Updates, or extends? If the former, the draft needs an “Updates RFC > 5334” field in the initial heading. Added the field. A complete diff of the changes since -09 so far is attached (and, as always, can be found in the Opus git repository: ). I'll post again when I finish reviewing the changes for normative language for generic Ogg requirements and the extra copyright grant. --------------090501040108020500070205 Content-Type: text/x-patch; name="changes-since-09.patch" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="changes-since-09.patch" diff --git a/doc/draft-ietf-codec-oggopus.xml b/doc/draft-ietf-codec-oggopus.xml index 124d488..b95e5bc 100644 --- a/doc/draft-ietf-codec-oggopus.xml +++ b/doc/draft-ietf-codec-oggopus.xml @@ -7,17 +7,18 @@ ]> - + Ogg Encapsulation for the Opus Audio Codec Mozilla Corporation
650 Castro Street @@ -100,18 +101,18 @@ Each page is associated with a particular logical stream and contains a capture pattern and checksum, flags to mark the beginning and end of the logical stream, and a 'granule position' that represents an absolute position in the stream, to aid seeking. A single page can contain up to 65,025 octets of packet data from up to 255 different packets. Packets can be split arbitrarily across pages, and continued from one page to the next (allowing packets much larger than would fit on a single page). Each page contains 'lacing values' that indicate how the data is partitioned - into packets, allowing a demuxer to recover the packet boundaries without - examining the encoded data. + into packets, allowing a demultiplexer (demuxer) to recover the packet + boundaries without examining the encoded data. A packet is said to 'complete' on a page when the page contains the final lacing value corresponding to that packet. This encapsulation defines the contents of the packet data, including the necessary headers, the organization of those packets into a logical stream, and the interpretation of the codec-specific granule position field. It does not attempt to describe or specify the existing Ogg container format. @@ -123,24 +124,16 @@ Readers unfamiliar with the basic concepts mentioned above are encouraged to
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "NOT RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in . - -Implementations that fail to satisfy one or more "MUST" requirements are - considered non-compliant. -Implementations that satisfy all "MUST" requirements, but fail to satisfy one - or more "SHOULD" requirements are said to be "conditionally compliant". -All other implementations are "unconditionally compliant". - -
An Ogg Opus stream is organized as follows. There are two mandatory header packets. @@ -175,25 +168,28 @@ The first (N - 1) Opus packets, if any, are packed one after another into the Ogg packet, using the self-delimiting framing from Appendix B of . The remaining Opus packet is packed at the end of the Ogg packet using the regular, undelimited framing from Section 3 of . All of the Opus packets in a single Ogg packet MUST be constrained to have the same duration. An implementation of this specification SHOULD treat any Opus packet whose duration is different from that of the first Opus packet in an Ogg packet as - if it were a malformed Opus packet with an invalid TOC sequence. + if it were a malformed Opus packet with an invalid Table Of Contents (TOC) + sequence. -The coding mode (SILK, Hybrid, or CELT), audio bandwidth, channel count, - duration (frame size), and number of frames per packet, are indicated in the - TOC (table of contents) sequence at the beginning of each Opus packet, as - described in Section 3.1 of . -The combination of mode, audio bandwidth, and frame size is referred to as - the configuration of an Opus packet. +The TOC sequence at the beginning of each Opus packet indicates the coding + mode, audio bandwidth, channel count, duration (frame size), and number of + frames per packet, as described in Section 3.1 + of . +The coding mode is one of SILK, Hybrid, or Constrained Energy Lapped Transform + (CELT). +The combination of coding mode, audio bandwidth, and frame size is referred to + as the configuration of an Opus packet. The first audio data page SHOULD NOT have the 'continued packet' flag set (which would indicate the first audio data packet is continued from a previous page). Packets MUST be placed into Ogg pages in order until the end of stream. Audio packets MAY span page boundaries. An implementation MUST treat a zero-octet audio data packet as if it were a @@ -264,18 +260,19 @@ All other pages with completed packets after the first MUST have a granule This guarantees that a demuxer can assign individual packets the same granule position when working forwards as when working backwards. For this to work, there cannot be any gaps.
In order to support capturing a real-time stream that has lost or not - transmitted packets, a muxer SHOULD emit packets that explicitly request the - use of Packet Loss Concealment (PLC) in place of the missing packets. + transmitted packets, a multiplexer (muxer) SHOULD emit packets that explicitly + request the use of Packet Loss Concealment (PLC) in place of the missing + packets. Implementations that fail to do so still MUST NOT increment the granule position for a page by anything other than the number of samples contained in packets that actually complete on that page. Only gaps that are a multiple of 2.5 ms are repairable, as these are the only durations that can be created by packet loss or discontinuous transmission. @@ -374,21 +371,21 @@ Therefore, the first few samples produced by the decoder do not correspond to These samples need to be stored and decoded, as Opus is an asymptotically convergent predictive codec, meaning the decoded contents of each frame depend on the recent history of decoder inputs. However, a player will want to skip these samples after decoding them. A 'pre-skip' field in the ID header (see ) signals - the number of samples which SHOULD be skipped (decoded but discarded) at the + the number of samples that SHOULD be skipped (decoded but discarded) at the beginning of the stream. This amount need not be a multiple of 2.5 ms, MAY be smaller than a single packet, or MAY span the contents of several packets. -These samples are not valid audio, and SHOULD NOT be played. +These samples are not valid audio. For example, if the first Opus frame uses the CELT mode, it will always produce 120 samples of windowed overlap-add data. However, the overlap data is initially all zeros (since there is no prior frame), meaning this cannot, in general, accurately represent the original audio. @@ -639,16 +636,17 @@ This is the number of samples (at 48 kHz) to discard from the decoder When cropping the beginning of existing Ogg Opus streams, a pre-skip of at least 3,840 samples (80 ms) is RECOMMENDED to ensure complete convergence in the decoder. Input Sample Rate (32 bits, unsigned, little endian): +This is the sample rate of the original input (before encoding), in Hz. This field is not the sample rate to use for playback of the encoded data. Opus can switch between internal audio bandwidths of 4, 6, 8, 12, and 20 kHz. Each packet in the stream can have a different audio bandwidth. Regardless of the audio bandwidth, the reference decoder supports decoding any stream at a sample rate of 8, 12, 16, 24, or 48 kHz. @@ -696,17 +694,17 @@ It is 20*log10 of the factor by which to scale the decoder output to achieve To apply the gain, an implementation could use
where output_gain is the raw 16-bit value from the header. -Virtually all players and media frameworks SHOULD apply it by default. +Players and media frameworks SHOULD apply it by default. If a player chooses to apply any volume adjustment or gain modification, such as the R128_TRACK_GAIN (see ), the adjustment MUST be applied in addition to this output gain in order to achieve playback at the normalized volume. A muxer SHOULD set this field to zero, and instead apply any gain prior to encoding, when this is possible and does not conflict with the user's wishes. A nonzero output gain indicates the gain was adjusted after encoding, or that @@ -720,36 +718,34 @@ The large range serves in part to ensure that gain can always be losslessly transferred between OpusHead and R128 gain tags (see below) without saturating. Channel Mapping Family (8 bits, unsigned): This octet indicates the order and semantic meaning of the output channels. -Each possible value of this octet indicates a mapping family, which defines a - set of allowed channel counts, and the ordered set of channel names for each - allowed channel count. +Each currently specified value of this octet indicates a mapping family, which + defines a set of allowed channel counts, and the ordered set of channel names + for each allowed channel count. The details are described in . Channel Mapping Table: This table defines the mapping from encoded streams to output channels. -It MUST be omitted when the channel mapping family is 0, but is - REQUIRED otherwise. Its contents are specified in . All fields in the ID headers are REQUIRED, except for the channel mapping table, which MUST be omitted when the channel mapping family is 0, but is REQUIRED otherwise. -Implementations SHOULD reject ID headers which do not contain enough data for - these fields, even if they contain a valid Magic Signature. +Implementations SHOULD reject streams with ID headers that do not contain + enough data for these fields, even if they contain a valid Magic Signature. Future versions of this specification, even backwards-compatible versions, might include additional fields in the ID header. If an ID header has a compatible major version, but a larger minor version, an implementation MUST NOT reject it for containing additional data not specified here, provided it still completes on the first page.
@@ -869,17 +865,17 @@ Special mapping: This channel mapping value also When the 'channel mapping family' octet has this value, the channel mapping table MUST be omitted from the ID header packet.
Allowed numbers of channels: 1...8. -Vorbis channel order. +Vorbis channel order (see below). Each channel is assigned to a speaker location in a conventional surround arrangement. Specific locations depend on the number of channels, and are given below in order of the corresponding channel indices. 1 channel: monophonic (mono). @@ -892,17 +888,17 @@ Specific locations depend on the number of channels, and are given below 8 channels: 7.1 surround (front left, front center, front right, side left, side right, rear left, rear right, LFE) This set of surround options and speaker location orderings is the same as those used by the Vorbis codec . The ordering is different from the one used by the WAVE and - FLAC formats, + Free Lossless Audio Codec (FLAC) formats, so correct ordering requires permutation of the output channels when decoding to or encoding from those formats. 'LFE' here refers to a Low Frequency Effects channel, often mapped to a subwoofer with no particular spatial position. Implementations SHOULD identify 'side' or 'rear' speaker locations with 'surround' and 'back' as appropriate when interfacing with audio formats or systems which prefer that terminology. @@ -924,18 +920,18 @@ Implementations SHOULD NOT produce output for channels mapped to stream index non-silent channels.
The remaining channel mapping families (2...254) are reserved. -An implementation encountering a reserved channel mapping family value SHOULD - act as though the value is 255. +A demuxer implementation encountering a reserved channel mapping family value + SHOULD act as though the value is 255.
An Ogg Opus player MUST support any valid channel mapping with a channel mapping family of 0 or 1, even if the number of channels does not match the physically connected audio hardware. @@ -1188,17 +1184,17 @@ If the least-significant bit of the first byte of this data is 1, then editors SHOULD preserve the contents of this data when updating the tags, but if this bit is 0, all such data MAY be treated as padding, and truncated or discarded as desired. The comment header can be arbitrarily large and might be spread over a large number of Ogg pages. -Implementations SHOULD avoid attempting to allocate excessive amounts of memory +Implementations MUST avoid attempting to allocate excessive amounts of memory when presented with a very large comment header. To accomplish this, implementations MAY reject a comment header larger than 125,829,120 octets, and MAY ignore individual comments that are not fully contained within the first 61,440 octets of the comment header.
@@ -1233,35 +1229,35 @@ R128_ALBUM_GAIN=111 representing the volume shift needed to normalize the overall volume when played as part of a particular collection of tracks. The gain is also a Q7.8 fixed point number in dB, as in the ID header's 'output gain' field. -An Ogg Opus stream MUST NOT have more than one of each tag, and if present - their values MUST be an integer from -32768 to 32767, inclusive, +An Ogg Opus stream MUST NOT have more than one of each of these tags, and if + present their values MUST be an integer from -32768 to 32767, inclusive, represented in ASCII as a base 10 number with no whitespace. A leading '+' or '-' character is valid. Leading zeros are also permitted, but the value MUST be represented by no more than 6 characters. Other non-digit characters MUST NOT be present. If present, R128_TRACK_GAIN and R128_ALBUM_GAIN MUST correctly represent the R128 normalization gain relative to the 'output gain' field specified in the ID header. If a player chooses to make use of the R128_TRACK_GAIN tag or the R128_ALBUM_GAIN tag, it MUST apply those gains in addition to the 'output gain' value. If a tool modifies the ID header's 'output gain' field, it MUST also update or remove the R128_TRACK_GAIN and R128_ALBUM_GAIN comment tags if present. -A muxer SHOULD assume that by default tools will respect the 'output gain' - field, and not the comment tag. +A muxer SHOULD place the gain it wants other tools to use by default into the + 'output gain' field, and not the comment tag. To avoid confusion with multiple normalization schemes, an Opus comment header SHOULD NOT contain any of the REPLAYGAIN_TRACK_GAIN, REPLAYGAIN_TRACK_PEAK, REPLAYGAIN_ALBUM_GAIN, or REPLAYGAIN_ALBUM_PEAK tags. normalization is preferred to the earlier REPLAYGAIN schemes because of its clear definition and adoption by industry. Peak normalizations are difficult to calculate reliably for lossy codecs @@ -1277,20 +1273,21 @@ In the authors' investigations they were not applied consistently or broadly
Technically, valid Opus packets can be arbitrarily large due to the padding format, although the amount of non-padding data they can contain is bounded. These packets might be spread over a similarly enormous number of Ogg pages. When encoding, implementations SHOULD limit the use of padding in audio data packets to no more than is necessary to make a variable bitrate (VBR) stream constant bitrate (CBR). -Demuxers SHOULD reject audio data packets larger than 61,440 octets per +Demuxers SHOULD reject audio data packets (treat them as if they were malformed + Opus packets with an invalid TOC sequence) larger than 61,440 octets per Opus stream. Such packets necessarily contain more padding than needed for this purpose. -Demuxers SHOULD avoid attempting to allocate excessive amounts of memory when +Demuxers MUST avoid attempting to allocate excessive amounts of memory when presented with a very large packet. Demuxers MAY reject or partially process audio data packets larger than 61,440 octets in an Ogg Opus stream with channel mapping families 0 or 1. Demuxers MAY reject or partially process audio data packets in any Ogg Opus stream if the packet is larger than 61,440 octets and also larger than 7,680 octets per Opus stream. The presence of an extremely large packet in the stream could indicate a @@ -1331,18 +1328,19 @@ This gives a size of 61,310 octets, which is rounded up to a multiple of
When encoding Opus streams, Ogg muxers SHOULD take into account the algorithmic delay of the Opus encoder. -In encoders derived from the reference implementation, the number of - samples can be queried with: +In encoders derived from the reference + implementation , the number of samples can be + queried with:
To achieve good quality in the very first samples of a stream, implementations @@ -1545,16 +1543,17 @@ The authors agree to grant third parties the irrevocable right to copy, use,
&rfc2119; &rfc3533; &rfc3629; + &rfc4732; &rfc5334; &rfc6381; &rfc6716; Loudness Recommendation EBU R128 @@ -1575,17 +1574,16 @@ The authors agree to grant third parties the irrevocable right to copy, use, - &rfc4732; &rfc6982; &rfc7587; FLAC - Free Lossless Audio Codec Format Description --------------090501040108020500070205-- From nobody Fri Dec 11 15:52:16 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id B87871A2119; Fri, 11 Dec 2015 15:52:15 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.6 X-Spam-Level: X-Spam-Status: No, score=-2.6 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7L7xoC6j-ffA; Fri, 11 Dec 2015 15:52:11 -0800 (PST) Received: from ipmail04.adl6.internode.on.net (ipmail04.adl6.internode.on.net [150.101.137.141]) by ietfa.amsl.com (Postfix) with ESMTP id E0E3E1A1F1D; Fri, 11 Dec 2015 15:52:10 -0800 (PST) Received: from ppp118-210-82-46.lns20.adl2.internode.on.net (HELO mailservice.shelbyville.oz) ([118.210.82.46]) by ipmail04.adl6.internode.on.net with ESMTP; 12 Dec 2015 10:22:09 +1030 Received: from localhost (localhost [127.0.0.1]) by mailservice.shelbyville.oz (Postfix) with ESMTP id A2E4EFFD84; Sat, 12 Dec 2015 10:22:08 +1030 (ACDT) X-Virus-Scanned: Debian amavisd-new at mailservice.shelbyville.oz Received: from mailservice.shelbyville.oz ([127.0.0.1]) by localhost (mailservice.shelbyville.oz [127.0.0.1]) (amavisd-new, port 10024) with LMTP id NBwqGjIB6PuM; Sat, 12 Dec 2015 10:22:03 +1030 (ACDT) Received: from hex.shelbyville.oz (hex.shelbyville.oz [192.168.1.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mailservice.shelbyville.oz (Postfix) with ESMTPS id D9964FF82B; Sat, 12 Dec 2015 10:22:03 +1030 (ACDT) Received: by hex.shelbyville.oz (Postfix, from userid 1000) id C0C3380470; Sat, 12 Dec 2015 10:22:03 +1030 (ACDT) Date: Sat, 12 Dec 2015 10:22:03 +1030 From: Ron To: "Timothy B. Terriberry" Message-ID: <20151211235203.GN17678@hex.shelbyville.oz> References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <566B4B47.9010809@xiph.org> User-Agent: Mutt/1.5.23 (2014-03-12) Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 11 Dec 2015 23:52:15 -0000 On Fri, Dec 11, 2015 at 02:16:39PM -0800, Timothy B. Terriberry wrote: > Ben Campbell wrote: > >Here is my AD evaluation of draft-ietf-codec-oggopus-09. Codec > >encapsulation is not my strong suit, so I suspect many of my comments > >can be answered with "this is the way we normally do this thing". But > >I'd still like to discuss these before going to IETF last call. > > Thank you for the very thorough review. Comments in-line below (as an > individual). Indeed. I think Tim has already covered most of my thoughts here, so just a couple of extra things: > >- 4.2, 2nd paragraph: > >-- "samples which SHOULD be skipped": Why not MUST? (also, s/which/that) > > Some specific applications might have a reason for looking at that data (for > example, when you have multiple files that you know were produced by > chopping up a single existing file without re-encoding). In practice we > haven't needed anything stronger to convince people to implement this > correctly (especially when we point them to > ). We might still want to clarify "These samples are not valid audio." at the end of that second paragraph - since as the last paragraph points out, some of them may actually be valid but just cropped for synchronisation or other aesthetic reasons. Maybe just moving that sentence up to the end of the first paragraph instead would make a bit more sense? Since it's the samples which are described there which are not valid audio, while the pre-skip field itself may overlap both those and some additional valid audio. > >- 5.1.1.4: Why not MUST? > > I think the idea was to allow assigning reserved values without explicitly > updating this specification. Even though I expect we _would_ want to update > this specification, given how long it's taken to get this draft published, > I'm wary of adding more pressure to people not to deploy new mappings > (because this RFC says they MUST not), even when it looks like they'll be > relatively stable and interoperable. At least not for the sole reason that > they haven't gotten an RFC number yet. Mark Harris and I had some discussion about this in the light of whether it should be MUST or SHOULD - and perhaps the most interesting thing to come out of that was wondering if we should in fact offer some advice on ways that people can safely use experimental new mappings before a new channel mapping number has been allocated for them. One option would be to suggest they use mapping 255 (which existing decoders SHOULD/MUST ignore) along with a special comment field of their own choosing which allows experimental decoders to know how to treat that stream. If the experiment fails, we lose an arbitrary unique name in the comment space, if it succeeds and gets consensus, we can add a mapping for it, and decoders can continue to recognise both the magic comment and the new mapping as appropriate. But we don't end up with incompatible uses for the same mapping number. It may be a bit late in the game to be adding something like that to this document - but we probably also don't strictly need to, since it doesn't change anything that is already normatively specified, it would just offer advice about how to proceed if you are developing a new mapping. Which means maybe it's also uncontroversial to add it too? > >- 13: > >Is this needed? Are the IETF BCP 78 and 79 provisions incorrect for this? > > This text was specifically requested by the Debian project (and I support it > personally). The same issue came up for RFC 6716 (for the same reasons). See > the discussion at > and > . > > Reviewing those conversations, I do notice that the language of that section > was changed after last call for RFC 6716, but the corresponding language in > our source repository was not updated (because it was added to the document > boilerplate manually by the RFC Editor, and not included in a separate > section). Yes, I think this probably needs to be changed to something more like what IETF legal suggested for RFC 6716, granting "the right to extract sections 1 - 14 ...". This allows people to make liberal use of the details in this document, but ensures they can't misrepresent any such use as somehow being an RFC. Cheers, Ron From nobody Mon Dec 14 14:56:32 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 74D4A1A1BC0; Mon, 14 Dec 2015 14:56:31 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.393 X-Spam-Level: X-Spam-Status: No, score=-2.393 tagged_above=-999 required=5 tests=[BAYES_20=-0.001, HELO_MISMATCH_ORG=0.611, HOST_MISMATCH_COM=0.311, MISSING_HEADERS=1.021, RCVD_IN_DNSWL_HI=-5, SPF_SOFTFAIL=0.665] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id eWIRnIckyl2A; Mon, 14 Dec 2015 14:56:29 -0800 (PST) Received: from smtp.mozilla.org (mx2.scl3.mozilla.com [63.245.214.156]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id A01171A02F1; Mon, 14 Dec 2015 14:56:29 -0800 (PST) Received: from localhost (localhost6.localdomain [127.0.0.1]) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTP id 51701C0A7F; Mon, 14 Dec 2015 22:56:29 +0000 (UTC) X-Virus-Scanned: amavisd-new at mozilla.org Received: from smtp.mozilla.org ([127.0.0.1]) by localhost (mx2.mail.scl3.mozilla.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id D-tzSMW9W8wu; Mon, 14 Dec 2015 22:56:29 +0000 (UTC) Received: from [10.252.27.204] (corp.mtv2.mozilla.com [63.245.221.32]) (Authenticated sender: tterriberry@mozilla.com) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTPSA id 3A7D2C0599; Mon, 14 Dec 2015 22:56:29 +0000 (UTC) Message-ID: <566F491D.5020307@xiph.org> Date: Mon, 14 Dec 2015 14:56:29 -0800 From: "Timothy B. Terriberry" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:29.0) Gecko/20100101 SeaMonkey/2.26 MIME-Version: 1.0 References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <20151211235203.GN17678@hex.shelbyville.oz> In-Reply-To: <20151211235203.GN17678@hex.shelbyville.oz> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 14 Dec 2015 22:56:31 -0000 Ron wrote: > mapping. Which means maybe it's also uncontroversial to add it too? I think my preference would be to leave it out. We had many of these extension points in Vorbis (transform type, floor type, residue type, mapping type, window type), and there were never any conflicts. Having an explicit "255 == unidentified" mapping is already more machinery than Vorbis had. Let's not invent more complications we don't know we need. From nobody Fri Dec 18 16:19:40 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 1C00F1A0103; Fri, 18 Dec 2015 16:19:39 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.91 X-Spam-Level: X-Spam-Status: No, score=-1.91 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, T_RP_MATCHES_RCVD=-0.01] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 7BaMs8ruQmTv; Fri, 18 Dec 2015 16:19:34 -0800 (PST) Received: from nostrum.com (raven-v6.nostrum.com [IPv6:2001:470:d:1130::1]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 7BE211A0100; Fri, 18 Dec 2015 16:19:34 -0800 (PST) Received: from [10.0.1.10] (cpe-70-119-203-4.tx.res.rr.com [70.119.203.4]) (authenticated bits=0) by nostrum.com (8.15.2/8.14.9) with ESMTPSA id tBJ0JX6Q067051 (version=TLSv1 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO); Fri, 18 Dec 2015 18:19:33 -0600 (CST) (envelope-from ben@nostrum.com) X-Authentication-Warning: raven.nostrum.com: Host cpe-70-119-203-4.tx.res.rr.com [70.119.203.4] claimed to be [10.0.1.10] From: "Ben Campbell" To: "Timothy B. Terriberry" Date: Fri, 18 Dec 2015 18:19:32 -0600 Message-ID: <25D8812E-3CEF-41BB-A82D-1A4B0524F439@nostrum.com> In-Reply-To: <566B4B47.9010809@xiph.org> References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> MIME-Version: 1.0 Content-Type: text/plain; format=flowed Content-Transfer-Encoding: quoted-printable X-Mailer: MailMate (1.9.3r5187) Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Sat, 19 Dec 2015 00:19:39 -0000 Thanks for the response. A few more comments on the substantive section = inline. I will get to the editorial parts in a separate email. (I = removed sections that do not seem to need further discussion.) Thanks! Ben. On 11 Dec 2015, at 16:16, Timothy B. Terriberry wrote: [...] > >> - 3, last paragraph: >> -- "implementations need to be prepared": >> maybe that should be a MUST or SHOULD? > > I think the intent was just to point out that someone might violate = > the preceding SHOULD. It's not clear what actionable thing we'd be = > saying they MUST do (it would depend on the particular = > implementation). I guess you can say MUST gracefully handle... That makes me wonder why the end-of-stream flag part is a SHOULD and not = a MUST, though. Are there reasonable circumstances to violate that = SHOULD? > >> -- "last page SHOULD NOT be a continued packet": >> Why not MUST NOT? Can you describe reasons someone might reasonably >> violate this? > > The intent was to allow but discourage page-level editing of files = > (i.e., cutting and splicing files by copying complete pages without = > examining the packets on them). Even just saving a live stream that = > drops at some point, you might reasonably expect to just be able to = > copy the data you received to a file. A MUST here would mean you would = > have to parse the pages, identify the continued packets from the = > lacing values, remove the extraneous data from the last page if you = > found some, repack the page buffer, and recompute the page checksum. = > That's a lot more machinery, and people would likely just ignore the = > requirement in practice. See also the discussion at = > . As stated, it still requires that extra work, unless the implementor has = a really good reason and fully understands the consequences. In this = case, it seems to mean SHOULD do that work unless you don't want to. I guess this really depends on how strongly you want to discourage the = behavior, and what breaks if people ignore that discouragement. This = might be better stated as non-normative guidance. Given that an = implementation has to be able to deal with a continued packet last page = anyway, does it really matter? If you want to keep it as a SHOULD NOT, it would be helpful to mention = the reasonable violations you envision in the text. > >> - 4.2, 2nd paragraph: >> -- "samples which SHOULD be skipped": Why not MUST? (also, = >> s/which/that) > > Some specific applications might have a reason for looking at that = > data (for example, when you have multiple files that you know were = > produced by chopping up a single existing file without re-encoding). It would be helpful to mention that in the text. > In practice we haven't needed anything stronger to convince people to = > implement this correctly (especially when we point them to = > ). I'm not don't think that SHOULD vs MUST logic is contemplated by RFC = 2119 :-) It's not so much how big of a stick you need to get people to = do the right thing as it is how bad do things break if people do the = wrong thing. (I do agree with the sentiment to not put in MUSTs that you = know people won't pay attention to, but I don't think thats the case = here.) [...] > >> - 4.3: Should the reference to [vorbis-trim] be normative? Can the >> feature be implemented without understanding the referenced doc? > > The answer to the second question is "yes", which I think makes the = > answer to the first question "no". The reason this is here is to tell = > people familiar with how Vorbis works what mechanism they should use = > to achieve the same outcome in Opus. But if you're not familiar with = > Vorbis, then you can safely ignore the whole paragraph. Okay. > >> - 4.6: Should the reference to [seeking] be normative? Can the = >> section >> be implemented without understanding it? > > It is possible to implement seeking without knowing anything other = > than what is described in RFC 3533, but the linked documentation does = > help considerably. > > I would argue that the section itself is not normative, in that you = > could implement seeking with some other method (for example, linearly = > scanning the whole file on open to build an index, possibly caching = > such indexes for files that are opened frequently: this would not be = > my recommended approach, but people have done it in practice). You = > could also choose to implement only approximate seeking by just = > guessing a file position and decoding from there (also done in = > practice by some players, and a common method for other formats like = > MP3). Or you could choose not to implement seeking at all. The current wording says "see [seeking] for general implementation = guidelines." If those are optional guidelines, it would help to say so, = or describe it as "some example guidelines". OTOH, if you really prefer = the approach in [seeking], then it still makes sense for it to be = normative. A reference required to understand and implement an optional feature = should still be normative. [...] > >> -- "Implementations SHOULD reject ID headers" >> What does it mean, in context, to "reject" headers? > > The same thing as rejecting a stream (i.e., don't try to play/decode = > it). Changed to say that instead. Which "that"? Maybe this is a conventional way to say things in the codec community, = but when I see "reject" I usually think that means you generate some = sort of error back to the source of the thing you reject. (As opposed to = "ignore") > >> - 5.1.1.4: Why not MUST? > > I think the idea was to allow assigning reserved values without = > explicitly updating this specification. Even though I expect we = > _would_ want to update this specification, given how long it's taken = > to get this draft published, I'm wary of adding more pressure to = > people not to deploy new mappings (because this RFC says they MUST = > not), even when it looks like they'll be relatively stable and = > interoperable. At least not for the sole reason that they haven't = > gotten an RFC number yet. If this spec reserves those values, then people still need to update it = to assign them. If you want to make it extensible, then perhaps it needs = an IANA registry, which, depending on the assignment policy, may not = require an RFC at all (seems like "expert review" or "specification = required" might make sense). Or if you want people to be able to play with other mappings without = "officially" assigning values, you could make some or all of the range = "for experimental use". > >> - 5.2, last two paragraph: Why are the two SHOULDs not MUSTs? > > For the first SHOULD, this is the same as above: some other = > specification may eventually define the format of the data here and = > how to modify it, and I didn't think we'd want to make people = > supporting that specification in violation of this specification. I'm = > also leaning towards keeping this one a SHOULD since a) the eventual = > format of this data may apply to all codecs which share the = > vorbiscomment style of metadata, and b) we would probably try to do = > this informally first to gain implementation experience, and only = > bring it to the IETF if there is sufficient interest. I didn't want to = > make updating this document a prerequisite for that. See also the = > discussion in this thread: = > = > (skip to = > = > for the resolution of the debate). I don't think the fact that future specs might change something is an = appropriate reason to pick SHOULD vs MUST. That's _always_ true, and = it's what our "UPDATES" process is for. The idea that you might want to = informally experiment with different approaches is more convincing. But = if that is the case, it would be helpful to have a sentence to that = effect in the text. > > For the second SHOULD, it felt like the right strength for a judgment = > call like "excessive". I have no objection to changing it to MUST. I think the consequences of not following the advice is the important = point. It sounds like this could be important for DoS attack avoidance, = and maybe even for buffer-overrun type things. Those fall into the = category of Bad Things. (I do agree "Excessive" is a bit vague, so if there were a way to state = this more concretely it would improve things. But that's true for a = SHOULD as much as for a MUST). [...] >> - 5.2.1, last paragraph: Why is the SHOULD NOT not MUST NOT? > > In some contexts, there may be no confusion with multiple = > normalization schemes. The language in this section was debated a good = > deal on the list (see threads = > , = > , = > and = > ). = > I believe everyone is satisfied with the current compromise, and would = > prefer not to change the strength of any normative statements in it to = > avoid re-igniting the debate. I agree that SHOULD NOT may be appropriate here. My point is the text = should explain why. Almost anytime a SHOULD occurs, it's helpful to = include text describing why a MUST is not appropriate. (Usually in the = form of examples or reasons the authors think would be appropriate to = violate the SHOULD). > >> -6, first paragraph: >> Why are the two SHOULDs not MUSTs? What does it mean to reject a = >> packet? > > For the first SHOULD, the fact that most decoders are likely to reject = > such packets is enough to discourage encoders from emitting them, and = > I was wary of imposing a MUST restriction when RFC 6716 chose not to = > do so. Did 6716 use a SHOULD? I assume that any 2119 keywords in this draft are intended to further = restrict things from 6716. If it's already covered normatively there, = then it doesn't need 2119 keywords here (unless directly quoted or = attributed). But in any case, I'm looking for justification in the text. = (I do consider preventing the use of excessive bandwidth a good enough = reason for a MUST > > For the second, this is the same as the previous comment about = > excessive allocations. I have no objection to changing this one, = > either. I realize now there were 3 SHOULDs in that paragraph. I mean the one = about avoiding excessive allocations. It sounds like [...] > >> [linear-predection] be normative? Can this be implemented without >> understanding that? If it does need to be normative, can you find a = >> more >> stable reference? > > Linear prediction is something that would be understood by "a person = > having ordinary skill in the art" (to borrow a phrase). We just wanted = > to include a reference to help those who might not be DSP engineers = > but wanted to learn more. Please consider saying something like "For more information on linear = prediction, see [XXX]." As it's currently stated, it looks like it says = MAY do X, where the reference specified X, which would require a = normative reference. [...] > >> - 13: >> Is this needed? Are the IETF BCP 78 and 79 provisions incorrect for = >> this? > > This text was specifically requested by the Debian project (and I = > support it personally). The same issue came up for RFC 6716 (for the = > same reasons). See the discussion at = > = > and = > . > > Reviewing those conversations, I do notice that the language of that = > section was changed after last call for RFC 6716, but the = > corresponding language in our source repository was not updated = > (because it was added to the document boilerplate manually by the RFC = > Editor, and not included in a separate section). That language is a bit of a problem for a standards track RFC. If people = are really attached to it, we can pursue allowing it. But I expect = objections down the line. Keep in mind "NOTE WELL" says that any = contributions are subject to the rules of BCP 78 and 79. In any case, 6716 is a bit more code centric , so the discussion made = more sense there. From nobody Mon Dec 21 13:31:11 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A1CE61ACD89; Mon, 21 Dec 2015 13:31:10 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -0.011 X-Spam-Level: X-Spam-Status: No, score=-0.011 tagged_above=-999 required=5 tests=[BAYES_40=-0.001, T_RP_MATCHES_RCVD=-0.01] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id aLmkD6l9uOzu; Mon, 21 Dec 2015 13:31:09 -0800 (PST) Received: from nostrum.com (raven-v6.nostrum.com [IPv6:2001:470:d:1130::1]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 7DB761ACD81; Mon, 21 Dec 2015 13:31:09 -0800 (PST) Received: from [10.0.1.10] (cpe-70-119-203-4.tx.res.rr.com [70.119.203.4]) (authenticated bits=0) by nostrum.com (8.15.2/8.14.9) with ESMTPSA id tBLLV8hA075635 (version=TLSv1 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO); Mon, 21 Dec 2015 15:31:08 -0600 (CST) (envelope-from ben@nostrum.com) X-Authentication-Warning: raven.nostrum.com: Host cpe-70-119-203-4.tx.res.rr.com [70.119.203.4] claimed to be [10.0.1.10] From: "Ben Campbell" To: "Timothy B. Terriberry" Date: Mon, 21 Dec 2015 15:31:07 -0600 Message-ID: <02FE33D2-476B-4CD5-927C-63BC3D4D4D25@nostrum.com> In-Reply-To: <566B4B47.9010809@xiph.org> References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit X-Mailer: MailMate (1.9.3r5187) Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Dec 2015 21:31:10 -0000 Hi, Here are my comments to your responses to my editorial comments. As before, I removed sections that do not seem to need further comment. Thanks! Ben. On 11 Dec 2015, at 16:16, Timothy B. Terriberry wrote: [...] > Editorial >> ========= [...] > > >> - 5, steps 2 and 3: >> These steps in combination seem to say “ decode at the highest >> available >> supported rate that the hardware can handle.” If so, the wording >> could >> be simplified. > > > It is a little bit more complicated. There are two rates here: one > used for decoding, and one used for playback. The difference between > steps 2 and 3 is that in step 3, the hardware's highest available > sample rate is not one of the rates supported by Opus decoding. For > example, if the highest available sample rate the hardware can handle > is 32 kHz, then this is not a "supported rate" (i.e., not supported by > Opus decoding), so step 2 does not apply and step 3 does. Step 3 says > to decode at the next highest "supported rate", which would be 48 kHz, > and then resample to a rate the hardware can handle (in this case 32 > kHz). > > I don't think that's equivalent to your summary of the two steps. I think I see. The point is that you decode rate still does not match the hardware rate, thus the resampling? OTOH, I now find myself confused by "next highest supported rate above this". Does "this" refer to the "hardware’s highest available sample rate"? So the decode rate is always equal to or higher than the hardware playback rate? [...] > > >> -10, last paragraph: >> >> Updates, or extends? If the former, the draft needs an “Updates RFC >> 5334” field in the initial heading. I didn't see a response to this one. From nobody Mon Dec 21 13:43:37 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 88EFF1ACDA0; Mon, 21 Dec 2015 13:43:36 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -3.414 X-Spam-Level: X-Spam-Status: No, score=-3.414 tagged_above=-999 required=5 tests=[BAYES_20=-0.001, HELO_MISMATCH_ORG=0.611, HOST_MISMATCH_COM=0.311, RCVD_IN_DNSWL_HI=-5, SPF_SOFTFAIL=0.665] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id dutk-_cIchna; Mon, 21 Dec 2015 13:43:35 -0800 (PST) Received: from smtp.mozilla.org (mx2.scl3.mozilla.com [63.245.214.156]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 2F70E1ACD9F; Mon, 21 Dec 2015 13:43:35 -0800 (PST) Received: from localhost (localhost6.localdomain [127.0.0.1]) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTP id A4F42C21D4; Mon, 21 Dec 2015 21:43:34 +0000 (UTC) X-Virus-Scanned: amavisd-new at mozilla.org Received: from smtp.mozilla.org ([127.0.0.1]) by localhost (mx2.mail.scl3.mozilla.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id HTLgG5ES2YOa; Mon, 21 Dec 2015 21:43:34 +0000 (UTC) Received: from [192.168.11.11] (pool-71-120-21-208.washdc.east.verizon.net [71.120.21.208]) (Authenticated sender: tterriberry@mozilla.com) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTPSA id ECA76C112D; Mon, 21 Dec 2015 21:43:33 +0000 (UTC) Message-ID: <56787285.7010803@xiph.org> Date: Mon, 21 Dec 2015 13:43:33 -0800 From: "Timothy B. Terriberry" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:29.0) Gecko/20100101 SeaMonkey/2.26 MIME-Version: 1.0 To: Ben Campbell References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <02FE33D2-476B-4CD5-927C-63BC3D4D4D25@nostrum.com> In-Reply-To: <02FE33D2-476B-4CD5-927C-63BC3D4D4D25@nostrum.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Dec 2015 21:43:36 -0000 A fast response because these are easy: Ben Campbell wrote: > I think I see. The point is that you decode rate still does not match > the hardware rate, thus the resampling? OTOH, I now find myself confused > by "next highest supported rate above this". Does "this" refer to the > "hardware’s highest available sample rate"? So the decode rate is always > equal to or higher than the hardware playback rate? Correct (except possibly if the hardware supports rates higher than 48 kHz, in which case step 4 might apply). >>> -10, last paragraph: >>> >>> Updates, or extends? If the former, the draft needs an “Updates RFC >>> 5334” field in the initial heading. > > I didn't see a response to this one. The response was: >> Added the field. Or, from the diff: >> - >> +> + updates="5334"> From nobody Mon Dec 21 14:14:35 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id ACF2C1ACDDE; Mon, 21 Dec 2015 14:14:34 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.91 X-Spam-Level: X-Spam-Status: No, score=-1.91 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, T_RP_MATCHES_RCVD=-0.01] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id vhNuET62jDqX; Mon, 21 Dec 2015 14:14:32 -0800 (PST) Received: from nostrum.com (raven-v6.nostrum.com [IPv6:2001:470:d:1130::1]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id DCE9F1ACDD2; Mon, 21 Dec 2015 14:14:32 -0800 (PST) Received: from [10.0.1.10] (cpe-70-119-203-4.tx.res.rr.com [70.119.203.4]) (authenticated bits=0) by nostrum.com (8.15.2/8.14.9) with ESMTPSA id tBLMERFS079663 (version=TLSv1 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO); Mon, 21 Dec 2015 16:14:28 -0600 (CST) (envelope-from ben@nostrum.com) X-Authentication-Warning: raven.nostrum.com: Host cpe-70-119-203-4.tx.res.rr.com [70.119.203.4] claimed to be [10.0.1.10] From: "Ben Campbell" To: Ron Date: Mon, 21 Dec 2015 16:14:27 -0600 Message-ID: <3D1E24B4-E9ED-42AF-A580-A024C25C3D2A@nostrum.com> In-Reply-To: <20151211235203.GN17678@hex.shelbyville.oz> References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <20151211235203.GN17678@hex.shelbyville.oz> MIME-Version: 1.0 Content-Type: text/plain; format=flowed X-Mailer: MailMate (1.9.3r5187) Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Dec 2015 22:14:34 -0000 On 11 Dec 2015, at 17:52, Ron wrote: [I have no opinion on the "not valid audio samples" discussion.] > >>> - 5.1.1.4: Why not MUST? >> >> I think the idea was to allow assigning reserved values without >> explicitly >> updating this specification. Even though I expect we _would_ want to >> update >> this specification, given how long it's taken to get this draft >> published, >> I'm wary of adding more pressure to people not to deploy new mappings >> (because this RFC says they MUST not), even when it looks like >> they'll be >> relatively stable and interoperable. At least not for the sole reason >> that >> they haven't gotten an RFC number yet. > > Mark Harris and I had some discussion about this in the light of > whether > it should be MUST or SHOULD - and perhaps the most interesting thing > to > come out of that was wondering if we should in fact offer some advice > on > ways that people can safely use experimental new mappings before a new > channel mapping number has been allocated for them. > > One option would be to suggest they use mapping 255 (which existing > decoders SHOULD/MUST ignore) along with a special comment field of > their own choosing which allows experimental decoders to know how to > treat that stream. It's a common practice to mark certain code points or ranges as "experimental" in IETF specs. > > If the experiment fails, we lose an arbitrary unique name in the > comment > space, if it succeeds and gets consensus, we can add a mapping for it, > and decoders can continue to recognise both the magic comment and the > new mapping as appropriate. But we don't end up with incompatible > uses > for the same mapping number. > > It may be a bit late in the game to be adding something like that to > this document - but we probably also don't strictly need to, since it > doesn't change anything that is already normatively specified, it > would > just offer advice about how to proceed if you are developing a new > mapping. Which means maybe it's also uncontroversial to add it too? I think adding that at this point would require mention on the working group list, to give people a chance to object. But that's not difficult. > > >>> - 13: >>> Is this needed? Are the IETF BCP 78 and 79 provisions incorrect for >>> this? >> >> This text was specifically requested by the Debian project (and I >> support it >> personally). The same issue came up for RFC 6716 (for the same >> reasons). See >> the discussion at >> >> and >> . >> >> Reviewing those conversations, I do notice that the language of that >> section >> was changed after last call for RFC 6716, but the corresponding >> language in >> our source repository was not updated (because it was added to the >> document >> boilerplate manually by the RFC Editor, and not included in a >> separate >> section). > > Yes, I think this probably needs to be changed to something more like > what IETF legal suggested for RFC 6716, granting "the right to extract > sections 1 - 14 ...". This allows people to make liberal use of the > details in this document, but ensures they can't misrepresent any such > use as somehow being an RFC. > Using the modified boilerplate from RFC 6716 would make more sense, if this document really needs it. But why does this document need it? It doesn't include code (normative or otherwise) like 6716. Or more to the point, why does this document need the extra grants more than any other standards-track RFC? Thanks! Ben. From nobody Mon Dec 21 14:21:14 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id C089E1ACDEA; Mon, 21 Dec 2015 14:21:12 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -1.91 X-Spam-Level: X-Spam-Status: No, score=-1.91 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, T_RP_MATCHES_RCVD=-0.01] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id fsoMCuth0L0P; Mon, 21 Dec 2015 14:21:11 -0800 (PST) Received: from nostrum.com (raven-v6.nostrum.com [IPv6:2001:470:d:1130::1]) (using TLSv1.2 with cipher ECDHE-RSA-AES128-GCM-SHA256 (128/128 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 4428A1ACDE7; Mon, 21 Dec 2015 14:21:11 -0800 (PST) Received: from [10.0.1.10] (cpe-70-119-203-4.tx.res.rr.com [70.119.203.4]) (authenticated bits=0) by nostrum.com (8.15.2/8.14.9) with ESMTPSA id tBLMLAWc080398 (version=TLSv1 cipher=DHE-RSA-AES128-SHA bits=128 verify=NO); Mon, 21 Dec 2015 16:21:10 -0600 (CST) (envelope-from ben@nostrum.com) X-Authentication-Warning: raven.nostrum.com: Host cpe-70-119-203-4.tx.res.rr.com [70.119.203.4] claimed to be [10.0.1.10] From: "Ben Campbell" To: "Timothy B. Terriberry" Date: Mon, 21 Dec 2015 16:21:10 -0600 Message-ID: <9BC52D4B-A133-4A54-8504-E88EEADB8D67@nostrum.com> In-Reply-To: <56787285.7010803@xiph.org> References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <02FE33D2-476B-4CD5-927C-63BC3D4D4D25@nostrum.com> <56787285.7010803@xiph.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8; format=flowed Content-Transfer-Encoding: 8bit X-Mailer: MailMate (1.9.3r5187) Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Dec 2015 22:21:12 -0000 On 21 Dec 2015, at 15:43, Timothy B. Terriberry wrote: > Ben Campbell wrote: > >> I think I see. The point is that you decode rate still does not match >> the hardware rate, thus the resampling? OTOH, I now find myself >> confused >> by "next highest supported rate above this". Does "this" refer to the >> "hardware’s highest available sample rate"? So the decode rate is >> always >> equal to or higher than the hardware playback rate? > > > Correct (except possibly if the hardware supports rates higher than 48 > kHz, in which case step 4 might apply). Got it. But I still think the "this" is ambiguous (between "highest hardware rate" and 48khz). So I propose: OLD: Otherwise, if the hardware’s highest available sample rate is less than 48 kHz, decode at the next highest supported rate above this and resample. NEW: Otherwise, if the hardware’s highest available sample rate is less than 48 kHz, decode at the next higher Opus supported rate above the hardware rate and resample to the hardware rate. END From nobody Mon Dec 21 14:45:04 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 081761ACE22; Mon, 21 Dec 2015 14:45:03 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -5.313 X-Spam-Level: X-Spam-Status: No, score=-5.313 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HELO_MISMATCH_ORG=0.611, HOST_MISMATCH_COM=0.311, RCVD_IN_DNSWL_HI=-5, SPF_SOFTFAIL=0.665] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id xVAJqHGbbGZ0; Mon, 21 Dec 2015 14:45:01 -0800 (PST) Received: from smtp.mozilla.org (mx2.scl3.mozilla.com [63.245.214.156]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id D4A1D1ACE20; Mon, 21 Dec 2015 14:45:01 -0800 (PST) Received: from localhost (localhost6.localdomain [127.0.0.1]) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTP id 4F34EC0F46; Mon, 21 Dec 2015 22:45:01 +0000 (UTC) X-Virus-Scanned: amavisd-new at mozilla.org Received: from smtp.mozilla.org ([127.0.0.1]) by localhost (mx2.mail.scl3.mozilla.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id sElwirauaLoZ; Mon, 21 Dec 2015 22:45:01 +0000 (UTC) Received: from [192.168.11.11] (pool-71-120-21-208.washdc.east.verizon.net [71.120.21.208]) (Authenticated sender: tterriberry@mozilla.com) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTPSA id B5E12C096D; Mon, 21 Dec 2015 22:45:00 +0000 (UTC) Message-ID: <567880EB.8000904@xiph.org> Date: Mon, 21 Dec 2015 14:44:59 -0800 From: "Timothy B. Terriberry" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:29.0) Gecko/20100101 SeaMonkey/2.26 MIME-Version: 1.0 To: Ben Campbell References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <02FE33D2-476B-4CD5-927C-63BC3D4D4D25@nostrum.com> <56787285.7010803@xiph.org> <9BC52D4B-A133-4A54-8504-E88EEADB8D67@nostrum.com> In-Reply-To: <9BC52D4B-A133-4A54-8504-E88EEADB8D67@nostrum.com> Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 8bit Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Dec 2015 22:45:03 -0000 Ben Campbell wrote: > OLD: > Otherwise, if the hardware’s highest available sample rate is less than > 48 kHz, decode at the next highest supported rate above this and resample. > NEW: > Otherwise, if the hardware’s highest available sample rate is less than > 48 kHz, decode at the next higher Opus supported rate above the hardware > rate and resample to the hardware rate. > END Okay, I've incorporated these changes into my local edits (though I used "above the highest available hardware rate" instead of just "above the hardware rate" so that it's extra clear we're talking about the same one). From nobody Tue Dec 22 12:32:07 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id A50CD1A8F44; Tue, 22 Dec 2015 12:32:06 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.6 X-Spam-Level: X-Spam-Status: No, score=-2.6 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, RCVD_IN_DNSWL_LOW=-0.7] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id O3lXoLITKMcA; Tue, 22 Dec 2015 12:32:03 -0800 (PST) Received: from ipmail07.adl2.internode.on.net (ipmail07.adl2.internode.on.net [150.101.137.131]) by ietfa.amsl.com (Postfix) with ESMTP id 2F35D1A1A91; Tue, 22 Dec 2015 12:32:03 -0800 (PST) Received: from ppp14-2-64-90.lns21.adl6.internode.on.net (HELO mailservice.shelbyville.oz) ([14.2.64.90]) by ipmail07.adl2.internode.on.net with ESMTP; 23 Dec 2015 07:02:02 +1030 Received: from localhost (localhost [127.0.0.1]) by mailservice.shelbyville.oz (Postfix) with ESMTP id D837CFFD97; Wed, 23 Dec 2015 07:01:49 +1030 (ACDT) X-Virus-Scanned: Debian amavisd-new at mailservice.shelbyville.oz Received: from mailservice.shelbyville.oz ([127.0.0.1]) by localhost (mailservice.shelbyville.oz [127.0.0.1]) (amavisd-new, port 10024) with LMTP id zuRKgOUMBXPW; Wed, 23 Dec 2015 07:01:47 +1030 (ACDT) Received: from hex.shelbyville.oz (hex.shelbyville.oz [192.168.1.6]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by mailservice.shelbyville.oz (Postfix) with ESMTPS id 85756FF88D; Wed, 23 Dec 2015 07:01:47 +1030 (ACDT) Received: by hex.shelbyville.oz (Postfix, from userid 1000) id 6D44780470; Wed, 23 Dec 2015 07:01:47 +1030 (ACDT) Date: Wed, 23 Dec 2015 07:01:47 +1030 From: Ron To: Ben Campbell Message-ID: <20151222203147.GY10797@hex.shelbyville.oz> References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <20151211235203.GN17678@hex.shelbyville.oz> <3D1E24B4-E9ED-42AF-A580-A024C25C3D2A@nostrum.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <3D1E24B4-E9ED-42AF-A580-A024C25C3D2A@nostrum.com> User-Agent: Mutt/1.5.23 (2014-03-12) Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 22 Dec 2015 20:32:06 -0000 On Mon, Dec 21, 2015 at 04:14:27PM -0600, Ben Campbell wrote: > On 11 Dec 2015, at 17:52, Ron wrote: > > > >>>- 5.1.1.4: Why not MUST? > >> > >>I think the idea was to allow assigning reserved values without > >>explicitly > >>updating this specification. Even though I expect we _would_ want to > >>update > >>this specification, given how long it's taken to get this draft > >>published, > >>I'm wary of adding more pressure to people not to deploy new mappings > >>(because this RFC says they MUST not), even when it looks like they'll > >>be > >>relatively stable and interoperable. At least not for the sole reason > >>that > >>they haven't gotten an RFC number yet. > > > >Mark Harris and I had some discussion about this in the light of whether > >it should be MUST or SHOULD - and perhaps the most interesting thing to > >come out of that was wondering if we should in fact offer some advice on > >ways that people can safely use experimental new mappings before a new > >channel mapping number has been allocated for them. > > > >One option would be to suggest they use mapping 255 (which existing > >decoders SHOULD/MUST ignore) along with a special comment field of > >their own choosing which allows experimental decoders to know how to > >treat that stream. > > It's a common practice to mark certain code points or ranges as > "experimental" in IETF specs. > > >If the experiment fails, we lose an arbitrary unique name in the comment > >space, if it succeeds and gets consensus, we can add a mapping for it, > >and decoders can continue to recognise both the magic comment and the > >new mapping as appropriate. But we don't end up with incompatible uses > >for the same mapping number. > > > >It may be a bit late in the game to be adding something like that to > >this document - but we probably also don't strictly need to, since it > >doesn't change anything that is already normatively specified, it would > >just offer advice about how to proceed if you are developing a new > >mapping. Which means maybe it's also uncontroversial to add it too? > > I think adding that at this point would require mention on the working group > list, to give people a chance to object. But that's not difficult. I don't feel a strong need to see this added, but I mentioned it precisely because it did seem like a discussion that should be aired on the list, if only as a reminder for later about why we didn't do this and a note that we did think about it. I think Tim was already looking into what we need to do to set up an IANA registry for this. Which is seeming like it's probably the right thing here. > >>>- 13: > >>>Is this needed? Are the IETF BCP 78 and 79 provisions incorrect for > >>>this? > >> > >>This text was specifically requested by the Debian project (and I > >>support it > >>personally). The same issue came up for RFC 6716 (for the same reasons). > >>See > >>the discussion at > >> and > >>. > >> > >>Reviewing those conversations, I do notice that the language of that > >>section > >>was changed after last call for RFC 6716, but the corresponding language > >>in > >>our source repository was not updated (because it was added to the > >>document > >>boilerplate manually by the RFC Editor, and not included in a separate > >>section). > > > >Yes, I think this probably needs to be changed to something more like > >what IETF legal suggested for RFC 6716, granting "the right to extract > >sections 1 - 14 ...". This allows people to make liberal use of the > >details in this document, but ensures they can't misrepresent any such > >use as somehow being an RFC. > > > > Using the modified boilerplate from RFC 6716 would make more sense, if this > document really needs it. But why does this document need it? It doesn't > include code (normative or otherwise) like 6716. Or more to the point, why > does this document need the extra grants more than any other standards-track > RFC? The crux here is twofold. We have organisations including, but not limited to, Debian, who cannot distribute this included with an implementation of it, unless it also grants people the right to redistribute potentially modified versions of it. A restriction which prevents people from claiming that a modified version is an RFC is fine, but people need to be able to cut selected parts out of it (possibly to include just selected parts in the documentation of their own library or applications), and/or adapt them for other purposes too. But beyond that, this document itself cribbed as much as we could from the work of other previous Ogg mappings, and includes things which are more generally useful, like the channel mapping definitions and the downmixing matrices, just to pick two. So it's not unreasonable to think this may in turn be a base which future work is built on, and in fact we'd encourage that, because it's definitely good not to change things that aren't broken and don't need changing for other future mappings. We'd also encourage that work to ultimately end up in an IETF working group too - but any initial work at least is quite likely to start outside of one until things are proven enough to take that next step. Ultimately, the rationale for this is actually almost identical to 6716, the only difference is 6716 created the weird situation where the normative part *was* the code, and it was already covered by a grant even more liberal than what we'd like for the descriptive text. The alternative (which we discussed with Robert Sparks for 6716) is the authors could publish a separate "this is not an RFC" version outside of the IETF with the extra grants - but that didn't seem like it would create less confusion than the solution IETF legal arrived at - which seems pretty close to ideal for any document that might be the basis for future or continuing work. We don't want people to be claiming that modified versions are an RFC, but we would like them to be able to reuse anything out of this that it makes sense to, as freely as we can allow them to. Otherwise we just artificially stifle further progress (or force everyone to turn a blind eye when people do things we didn't explicitly grant them the right to do - and that is the part I think we all share the desire to avoid). Cheers, Ron From nobody Mon Dec 28 05:00:45 2015 Return-Path: X-Original-To: codec@ietfa.amsl.com Delivered-To: codec@ietfa.amsl.com Received: from localhost (ietfa.amsl.com [127.0.0.1]) by ietfa.amsl.com (Postfix) with ESMTP id 9F9A51A90BD; Mon, 28 Dec 2015 05:00:43 -0800 (PST) X-Virus-Scanned: amavisd-new at amsl.com X-Spam-Flag: NO X-Spam-Score: -2.013 X-Spam-Level: X-Spam-Status: No, score=-2.013 tagged_above=-999 required=5 tests=[BAYES_50=0.8, HELO_MISMATCH_ORG=0.611, HOST_MISMATCH_COM=0.311, J_CHICKENPOX_81=0.6, RCVD_IN_DNSWL_HI=-5, SPF_SOFTFAIL=0.665] autolearn=ham Received: from mail.ietf.org ([4.31.198.44]) by localhost (ietfa.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id G7MU4kEwOjRZ; Mon, 28 Dec 2015 05:00:37 -0800 (PST) Received: from smtp.mozilla.org (mx2.scl3.mozilla.com [63.245.214.156]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by ietfa.amsl.com (Postfix) with ESMTPS id 5822A1A90BC; Mon, 28 Dec 2015 05:00:37 -0800 (PST) Received: from localhost (localhost6.localdomain [127.0.0.1]) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTP id F01A7C1510; Mon, 28 Dec 2015 13:00:36 +0000 (UTC) X-Virus-Scanned: amavisd-new at mozilla.org Received: from smtp.mozilla.org ([127.0.0.1]) by localhost (mx2.mail.scl3.mozilla.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id I_ryr-HOIF0b; Mon, 28 Dec 2015 13:00:35 +0000 (UTC) Received: from [10.0.0.42] (c-24-63-237-33.hsd1.ct.comcast.net [24.63.237.33]) (Authenticated sender: tterriberry@mozilla.com) by mx2.mail.scl3.mozilla.com (Postfix) with ESMTPSA id 9EDE9BFF00; Mon, 28 Dec 2015 13:00:34 +0000 (UTC) Message-ID: <56813271.10309@xiph.org> Date: Mon, 28 Dec 2015 05:00:33 -0800 From: "Timothy B. Terriberry" User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:29.0) Gecko/20100101 SeaMonkey/2.26 MIME-Version: 1.0 To: Ben Campbell References: <86ACD2D0-02B6-473E-9E35-B9980166D9A0@nostrum.com> <566B4B47.9010809@xiph.org> <25D8812E-3CEF-41BB-A82D-1A4B0524F439@nostrum.com> In-Reply-To: <25D8812E-3CEF-41BB-A82D-1A4B0524F439@nostrum.com> Content-Type: multipart/mixed; boundary="------------020004000201070508010308" Archived-At: Cc: codec@ietf.org, draft-ietf-codec-oggopus.all@ietf.org Subject: Re: [codec] AD Evaluation of draft-ietf-codec-oggopus-09 X-BeenThere: codec@ietf.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: Codec WG List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 28 Dec 2015 13:00:44 -0000 This is a multi-part message in MIME format. --------------020004000201070508010308 Content-Type: text/plain; charset=UTF-8; format=flowed Content-Transfer-Encoding: 7bit Ben Campbell wrote: > That makes me wonder why the end-of-stream flag part is a SHOULD and not > a MUST, though. Are there reasonable circumstances to violate that SHOULD? The usual case is again, saving a live stream that gets interrupted before it ends. > This might be better stated as non-normative guidance. Given that this is a general Ogg requirement, and not really Opus-specific, I think you are right. Here are the changes I propose: -The first audio data page SHOULD NOT have the 'continued packet' flag set - (which would indicate the first audio data packet is continued from a previous - page). -Packets MUST be placed into Ogg pages in order until the end of stream. -Audio packets MAY span page boundaries. +Packets are placed into Ogg pages in order until the end of stream. +Audio data packets might span page boundaries. +The first audio data page could have the 'continued packet' flag set + (indicating the first audio data packet is continued from a previous page) if, + for example, it was a live stream joined mid-broadcast, with the headers + pasted on the front. +A demuxer SHOULD NOT attempt to decode the data for the first packet on a page + with the 'continued packet' flag set if the previous page with packet data + does not end in a continued packet (i.e., did not end with a lacing value of + 255) or if the page sequence numbers are not consecutive, unless the demuxer + has some special knowledge that would allow it to interpret this data + despite the missing pieces. Along with -The last page SHOULD have the 'end of stream' flag set, but implementations - need to be prepared to deal with truncated streams that do not have a page - marked 'end of stream'. -The final packet on the last page SHOULD NOT be a continued packet, i.e., the - final lacing value SHOULD be less than 255. +A logical stream ends with a page with the 'end of stream' flag set, but + implementations need to be prepared to deal with truncated streams that do not + have a page marked 'end of stream'. +There is no reason for the final packet on the last page to be a continued + packet, i.e., for the final lacing value to be less than 255. +However, demuxers might encounter such streams, possibly as the result of a + transfer that did not complete or of corruption. +A demuxer SHOULD NOT attempt to decode the data from a packet that continues + onto a subsequent page (i.e., when the page ends with a lacing value of 255) + if the next page with packet data does not have the 'continued packet' flag + set or does not exist, or if the page sequence numbers are not consecutive, + unless the demuxer has some special knowledge that would allow it to interpret + this data despite the missing pieces. That gives specific guidance on what a demuxer needs to do to avoid corrupting the Opus state machine with garbage data (which is generally worse for Opus than simply skipping decode of a packet). It is also what is most broadly implemented. > It would be helpful to mention that in the text. I added something about that. > 2119 :-) It's not so much how big of a stick you need to get people to > do the right thing as it is how bad do things break if people do the > wrong thing. (I do agree with the sentiment to not put in MUSTs that you Given that historically, many audio formats (including MP3) did not support sample-accurate durations at all, I don't think the world ends. But it certainly would break features like gapless playback that some users find important. > The current wording says "see [seeking] for general implementation > guidelines." If those are optional guidelines, it would help to say so, > or describe it as "some example guidelines". OTOH, if you really prefer > the approach in [seeking], then it still makes sense for it to be > normative. I've added the optional/example qualifier. I also updated the claim on the number of bisections required, as our implementations have improved since the draft was first written. >>> What does it mean, in context, to "reject" headers? >> >> The same thing as rejecting a stream (i.e., don't try to play/decode >> it). Changed to say that instead. > > Which "that"? To quote the attached diff: -Implementations SHOULD reject ID headers which do not contain enough data for - these fields, even if they contain a valid Magic Signature. +Implementations SHOULD reject streams with ID headers that do not contain + enough data for these fields, even if they contain a valid Magic Signature. > Maybe this is a conventional way to say things in the codec community, > but when I see "reject" I usually think that means you generate some > sort of error back to the source of the thing you reject. (As opposed to > "ignore") Well, the most likely behavior here is to tell the user that the file is not a valid Ogg Opus file, so your sense of "generating some sort of error back" is in fact correct. But that makes some application-level assumptions I'm not sure I'm comfortable writing normative text around. Any ideas for a better way to phrase this? > If this spec reserves those values, then people still need to update it > to assign them. If you want to make it extensible, then perhaps it needs > an IANA registry, which, depending on the assignment policy, may not > require an RFC at all (seems like "expert review" or "specification > required" might make sense). An IANA registry actually does not seem like a terrible idea. Reading RFC 5226 it doesn't seem too hard to set one up. I attempted to draft some text. > it's what our "UPDATES" process is for. The idea that you might want to > informally experiment with different approaches is more convincing. But > if that is the case, it would be helpful to have a sentence to that > effect in the text. Okay, I will add one. > (I do agree "Excessive" is a bit vague, so if there were a way to state > this more concretely it would improve things. But that's true for a > SHOULD as much as for a MUST). The problem is that it depends very much on context. 60 kB might already be seen as excessive for an embedded system, but would never be noticed on a desktop. >> In some contexts, there may be no confusion with multiple >> normalization schemes. The language in this section was debated a good >> [...] > > I agree that SHOULD NOT may be appropriate here. My point is the text > should explain why. Almost anytime a SHOULD occurs, it's helpful to I agree with your point very much in principle. We just may not have been as careful about it in practice as we should have been (and I appreciate you forcing us to clarify our thinking now). I'll add something to the effect of what I said above. > Did 6716 use a SHOULD? It imposed no restrictions on the amount of padding at all. > attributed). But in any case, I'm looking for justification in the text. > (I do consider preventing the use of excessive bandwidth a good enough > reason for a MUST I attempted to add some. >> For the second, this is the same as the previous comment about >> excessive allocations. I have no objection to changing this one, either. > > I realize now there were 3 SHOULDs in that paragraph. I mean the one > about avoiding excessive allocations. It sounds like You may have forgotten to complete your thought here. > Please consider saying something like "For more information on linear > prediction, see [XXX]." As it's currently stated, it looks like it says > MAY do X, where the reference specified X, which would require a > normative reference. That's a good suggestion. Done. > That language is a bit of a problem for a standards track RFC. If people You mean the current text, or the actual text that wound up in RFC 6716? > are really attached to it, we can pursue allowing it. But I expect > objections down the line. Keep in mind "NOTE WELL" says that any > contributions are subject to the rules of BCP 78 and 79. Yes, I think this has more to do with letting people use such contributions _outside_ the context of the IETF (i.e., it grants additional rights to those already granted in BCP 78 and 79, which I understand is allowed). You are probably right about people objecting down the line (again), but I am happy to address those objections when they come, as we did for RFC 6716. > In any case, 6716 is a bit more code centric , so the discussion made > more sense there. Well, from our point of view, the XML source for both RFCs is included in the repository with libopus, which is distributed in Debian. If we didn't have some kind of grant like this, they would have to somehow strip those documents out. So it is not so much the code being in the RFC as the draft being distributed with the code. In any case, I have attempted to draft an RFC Editor's note requesting the same text that RFC 6716 used. Since that text grants a license from the IETF Trust, instead of the authors, I assume that someone from the Trust will have to approve of it, however (and reading it seems that the current opinion of the Trust is that this approach is legally required). I have also gone through and finished removing the 2119 language for general Ogg requirements. The diff of all changes is attached. If these look good (or if it's simply helps in reviewing them) I can publish a new version. --------------020004000201070508010308 Content-Type: text/x-patch; name="more-changes-since-09.patch" Content-Transfer-Encoding: 7bit Content-Disposition: attachment; filename="more-changes-since-09.patch" diff --git a/doc/draft-ietf-codec-oggopus.xml b/doc/draft-ietf-codec-oggopus.xml index b95e5bc..7489c20 100644 --- a/doc/draft-ietf-codec-oggopus.xml +++ b/doc/draft-ietf-codec-oggopus.xml @@ -1,14 +1,15 @@ + ]> @@ -135,19 +136,19 @@ The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", An Ogg Opus stream is organized as follows. There are two mandatory header packets. The first packet in the logical Ogg bitstream MUST contain the identification (ID) header, which uniquely identifies a stream as Opus audio. The format of this header is defined in . -It MUST be placed alone (without any other packet data) on the first page of - the logical Ogg bitstream, and MUST complete on that page. -This page MUST have its 'beginning of stream' flag set. +It is placed alone (without any other packet data) on the first page of + the logical Ogg bitstream, and completes on that page. +This page has its 'beginning of stream' flag set. The second packet in the logical Ogg bitstream MUST contain the comment header, which contains user-supplied metadata. The format of this header is defined in . It MAY span multiple pages, beginning on the second page of the logical stream. However many pages it spans, the comment header packet MUST finish the page on @@ -182,31 +183,46 @@ The TOC sequence at the beginning of each Opus packet indicates the coding frames per packet, as described in Section 3.1 of . The coding mode is one of SILK, Hybrid, or Constrained Energy Lapped Transform (CELT). The combination of coding mode, audio bandwidth, and frame size is referred to as the configuration of an Opus packet. -The first audio data page SHOULD NOT have the 'continued packet' flag set - (which would indicate the first audio data packet is continued from a previous - page). -Packets MUST be placed into Ogg pages in order until the end of stream. -Audio packets MAY span page boundaries. +Packets are placed into Ogg pages in order until the end of stream. +Audio data packets might span page boundaries. +The first audio data page could have the 'continued packet' flag set + (indicating the first audio data packet is continued from a previous page) if, + for example, it was a live stream joined mid-broadcast, with the headers + pasted on the front. +A demuxer SHOULD NOT attempt to decode the data for the first packet on a page + with the 'continued packet' flag set if the previous page with packet data + does not end in a continued packet (i.e., did not end with a lacing value of + 255) or if the page sequence numbers are not consecutive, unless the demuxer + has some special knowledge that would allow it to interpret this data + despite the missing pieces. An implementation MUST treat a zero-octet audio data packet as if it were a malformed Opus packet as described in Section 3.4 of . -The last page SHOULD have the 'end of stream' flag set, but implementations - need to be prepared to deal with truncated streams that do not have a page - marked 'end of stream'. -The final packet on the last page SHOULD NOT be a continued packet, i.e., the - final lacing value SHOULD be less than 255. +A logical stream ends with a page with the 'end of stream' flag set, but + implementations need to be prepared to deal with truncated streams that do not + have a page marked 'end of stream'. +There is no reason for the final packet on the last page to be a continued + packet, i.e., for the final lacing value to be less than 255. +However, demuxers might encounter such streams, possibly as the result of a + transfer that did not complete or of corruption. +A demuxer SHOULD NOT attempt to decode the data from a packet that continues + onto a subsequent page (i.e., when the page ends with a lacing value of 255) + if the next page with packet data does not have the 'continued packet' flag + set or does not exist, or if the page sequence numbers are not consecutive, + unless the demuxer has some special knowledge that would allow it to interpret + this data despite the missing pieces. There MUST NOT be any more pages in an Opus logical bitstream after a page marked 'end of stream'.
The granule position MUST be zero for the ID header page and the @@ -219,18 +235,18 @@ The granule position of an audio data page encodes the total number of PCM samples in the stream up to and including the last fully-decodable sample from the last packet completed on that page. The granule position of the first audio data page will usually be larger than zero, as described in . A page that is entirely spanned by a single packet (that completes on a - subsequent page) has no granule position, and the granule position field MUST - be set to the special value '-1' in two's complement. + subsequent page) has no granule position, and the granule position field is + set to the special value '-1' in two's complement. The granule position of an audio data page is in units of PCM audio samples at a fixed rate of 48 kHz (per channel; a stereo stream's granule position does not increment at twice the speed of a mono stream). It is possible to run an Opus decoder at other sampling rates, but the value in the granule position field always counts samples assuming a 48 kHz @@ -372,17 +388,18 @@ These samples need to be stored and decoded, as Opus is an asymptotically convergent predictive codec, meaning the decoded contents of each frame depend on the recent history of decoder inputs. However, a player will want to skip these samples after decoding them. A 'pre-skip' field in the ID header (see ) signals the number of samples that SHOULD be skipped (decoded but discarded) at the - beginning of the stream. + beginning of the stream, though some specific applications might have a reason + for looking at that data. This amount need not be a multiple of 2.5 ms, MAY be smaller than a single packet, or MAY span the contents of several packets. These samples are not valid audio. For example, if the first Opus frame uses the CELT mode, it will always produce 120 samples of windowed overlap-add data. @@ -520,19 +537,19 @@ Both of these will be greater than '0' in this case.
Seeking in Ogg files is best performed using a bisection search for a page whose granule position corresponds to a PCM position at or before the seek target. -With appropriately weighted bisection, accurate seeking can be performed with - just three or four bisections even in multi-gigabyte files. -See for general implementation guidance. +With appropriately weighted bisection, accurate seeking can be performed in + just one or two bisections on average, even in multi-gigabyte files. +See for an example of general implementation guidance. When seeking within an Ogg Opus stream, an implementation SHOULD start decoding (and discarding the output) at least 3840 samples (80 ms) prior to the seek target in order to ensure that the output audio is correct by the time it reaches the seek target. This 'pre-roll' is separate from, and unrelated to, the 'pre-skip' used at the @@ -655,18 +672,18 @@ The original sample rate of the audio passed to the encoder is not preserved An Ogg Opus player SHOULD select the playback sample rate according to the following procedure: If the hardware supports 48 kHz playback, decode at 48 kHz. Otherwise, if the hardware's highest available sample rate is a supported rate, decode at this sample rate. Otherwise, if the hardware's highest available sample rate is less than - 48 kHz, decode at the next highest supported rate above this and - resample. + 48 kHz, decode at the next higher Opus supported rate above the highest + available hardware rate and resample. Otherwise, decode at 48 kHz and resample. However, the 'Input Sample Rate' field allows the muxer to pass the sample rate of the original input stream as metadata. This is useful when the user requires the output sample rate to match the input sample rate. For example, when not playing the output, an implementation writing PCM format samples to disk might choose to resample the audio back to the original input @@ -1179,16 +1196,18 @@ Making this check before allocating the associated memory to contain the data Immediately following the user comment list, the comment header MAY contain zero-padding or other binary data which is not specified here. If the least-significant bit of the first byte of this data is 1, then editors SHOULD preserve the contents of this data when updating the tags, but if this bit is 0, all such data MAY be treated as padding, and truncated or discarded as desired. +This allows informal experimentation with the format of this binary data until + it can be specified later. The comment header can be arbitrarily large and might be spread over a large number of Ogg pages. Implementations MUST avoid attempting to allocate excessive amounts of memory when presented with a very large comment header. To accomplish this, implementations MAY reject a comment header larger than @@ -1252,17 +1271,18 @@ If a player chooses to make use of the R128_TRACK_GAIN tag or the If a tool modifies the ID header's 'output gain' field, it MUST also update or remove the R128_TRACK_GAIN and R128_ALBUM_GAIN comment tags if present. A muxer SHOULD place the gain it wants other tools to use by default into the 'output gain' field, and not the comment tag. To avoid confusion with multiple normalization schemes, an Opus comment header SHOULD NOT contain any of the REPLAYGAIN_TRACK_GAIN, REPLAYGAIN_TRACK_PEAK, - REPLAYGAIN_ALBUM_GAIN, or REPLAYGAIN_ALBUM_PEAK tags. + REPLAYGAIN_ALBUM_GAIN, or REPLAYGAIN_ALBUM_PEAK tags, unless they are only + to be used in some context where there is guaranteed to be no such confusion. normalization is preferred to the earlier REPLAYGAIN schemes because of its clear definition and adoption by industry. Peak normalizations are difficult to calculate reliably for lossy codecs because of variation in excursion heights due to decoder differences. In the authors' investigations they were not applied consistently or broadly enough to merit inclusion here.
@@ -1272,21 +1292,22 @@ In the authors' investigations they were not applied consistently or broadly
Technically, valid Opus packets can be arbitrarily large due to the padding format, although the amount of non-padding data they can contain is bounded. These packets might be spread over a similarly enormous number of Ogg pages. When encoding, implementations SHOULD limit the use of padding in audio data packets to no more than is necessary to make a variable bitrate (VBR) stream - constant bitrate (CBR). + constant bitrate (CBR), unless they have no reasonable way to determine what + is necessary. Demuxers SHOULD reject audio data packets (treat them as if they were malformed Opus packets with an invalid TOC sequence) larger than 61,440 octets per - Opus stream. -Such packets necessarily contain more padding than needed for this purpose. + Opus stream, unless they have a specific reason for allowing extra padding. +Such packets necessarily contain more padding than needed to make a stream CBR. Demuxers MUST avoid attempting to allocate excessive amounts of memory when presented with a very large packet. Demuxers MAY reject or partially process audio data packets larger than 61,440 octets in an Ogg Opus stream with channel mapping families 0 or 1. Demuxers MAY reject or partially process audio data packets in any Ogg Opus stream if the packet is larger than 61,440 octets and also larger than 7,680 octets per Opus stream. @@ -1339,20 +1360,21 @@ In encoders derived from the reference
To achieve good quality in the very first samples of a stream, implementations - MAY use linear predictive coding (LPC) extrapolation - to generate at least 120 extra samples at - the beginning to avoid the Opus encoder having to encode a discontinuous - signal. + MAY use linear predictive coding (LPC) extrapolation to generate at least 120 + extra samples at the beginning to avoid the Opus encoder having to encode a + discontinuous signal. +For more information on linear prediction, see + . For an input file containing 'length' samples, the implementation SHOULD set the pre-skip header value to (delay_samples + extra_samples), encode at least (length + delay_samples + extra_samples) samples, and set the granule position of the last page to (length + delay_samples + extra_samples). This ensures that the encoded file has the same duration as the original, with no time offset. The best way to pad the end of the stream is to also use LPC extrapolation, but zero-padding is also acceptable. @@ -1509,51 +1531,96 @@ In such cases the the '.opus' filename extension is NOT RECOMMENDED. In either case, this document updates to add 'opus' as a codecs parameter value with char[8]: 'OpusHead' as Codec Identifier.
-
+
This document updates the IANA Media Types registry to add .opus as a file extension for "audio/ogg", and to add itself as a reference alongside for "audio/ogg", "video/ogg", and "application/ogg" Media Types. + +This document defines a new registry "Opus Channel Mapping Families" to + indicate how the semantic meanings of the channels in a multi-channel Opus + stream are described. +IANA SHALL create a new name space of "Opus Channel Mapping Families". +All maintenance within and additions to the contents of this name space MUST be + according to the "Specification Requried with Expert Review" registration + policy as defined in . +Each registry entry consists of a Channel Mapping Family Number, which is + specified in decimal in the range 0 to 255, inclusive, and a Reference (or + list of references) +Each Reference must point to sufficient documentation to describe what + information is coded in the Opus identification header for this channel + mapping family, how a demuxer determines the Stream Count ('N') and Coupled + Stream Count ('M') from this information, and how it determines the proper + interpretation of each of the decoded channels. + + +This document defines three initial assignments for this registry. + + +ValueReference +0[RFCXXXX] +1[RFCXXXX] +255[RFCXXXX] + + +The designated expert will determine if the Reference points to a specification + that meets the requirements for permanence and ready availability laid out + in  and that it specifies the information + described above with sufficient clarity to allow interoperable + implementations. +
-Thanks to Mark Harris, Greg Maxwell, Christopher "Monty" Montgomery, and - Jean-Marc Valin for their valuable contributions to this document. +Thanks to Ben Campbell, Mark Harris, Greg Maxwell, Christopher "Monty" + Montgomery, Jean-Marc Valin, and Mo Zanaty for their valuable contributions to + this document. Additional thanks to Andrew D'Addesio, Greg Maxwell, and Vincent Penquerc'h for their feedback based on early implementations.
-
+
+ +In , "RFCXXXX" is to be replaced with the RFC number + assigned to this draft. + + +In the Copyright Notice at the start of the document, the following paragraph + is to be appended after the regular copyright notice text: + -The authors agree to grant third parties the irrevocable right to copy, use, - and distribute the work, with or without modification, in any medium, without - royalty, provided that, unless separate permission is granted, redistributed - modified works do not contain misleading author, version, name of work, or - endorsement information. +"The licenses granted by the IETF Trust to this RFC under Section 3.c of + the Trust Legal Provisions shall also include the right to extract text from + Sections 1 through 14 of this RFC and create derivative works from + these extracts, and to copy, publish, display, and distribute such derivative + works in any medium and for any purpose, provided that no such derivative work + shall be presented, displayed, or published in a manner that states or implies + that it is part of this RFC or any other IETF Document."
&rfc2119; &rfc3533; &rfc3629; &rfc4732; + &rfc5226; &rfc5334; &rfc6381; &rfc6716; Loudness Recommendation EBU R128 --------------020004000201070508010308--