Page 11, S2, para 3.
  Removed statement about differences between authors

Page 12.  top of page
  Fixed text wrapping problem

Page 13, S3.3, S3.4, Page 15 S3.8, Page 58 S15.1
  Changed BNF for conference-id, session-id, and option-tag to be more
  restrictive, and added the additional BNF to the spec:

  S3.3:
  conference-id  = 1*xchar
  
    Conference ids are externally generated, and often times expressable as
    URLs.  Though not specified to be a URL, recipients should be able to
    handle unmodified URLs.
  
  S3.4:
  session-id = 1*(ALPHA|DIGIT|safe)
  
    Session id's we want to be more restrictive with, since they are
    generated by RTSP implementations.  This ensures that they can be passed
    to other systems with minimal escaping necessary.
  
  S3.8:
    option-tag = 1*xchar
  
    Since many option tag proposals are URI-based, the complete set of URI
    characters is necessary to ensure that implementations can handle URIs if
    the standard evolves this way.
  
  S15.1:
  safe           = "$" | "-" | "_" | "." | "+"
  extra          = "!" | "*" | "'" | "(" | ")" | ","
  
  hex            = DIGIT | "A" | "B" | "C" | "D" | "E" | "F" |
		   "a" | "b" | "c" | "d" | "e" | "f"
  escape         = "%" hex hex
  
  reserved       = ";" | "/" | "?" | ":" | "@" | "&" | "="
  unreserved     = ALPHA | DIGIT | safe | extra
   
  xchar          = unreserved | reserved | escape

Page 7, S1.3
  Entity definition inserted from HTTP spec:

    Entity: The information transferred as the payload of a request or
    response. An entity consists of metainformation in the form of
    entity-header fields and content in the form of an entity-body, as
    described in Section 8.

  Also removed a couple uses of the word "entity" where it was inconsistant
  with the definition above.

Page 13, S3.5
  "---" in smpte-type BNF changed to "|" (.ps only)

Page 14, S3.6
  "---" in npt-time BNF changed to "|" (.ps only)

  npt-range added.

Page 15, S3.8.1, bullet 1
  "The name >>should<< not contain any spaces, ..."
  "should" changed to "MUST"

Page 18, S7.1.1, last para on page
  Fixed grammar on sentence starting "Note that RTSP adopts..." 

Page 23, para 4 "Each request..."
  The spec formerly said the sequence number is "not incremented".  The
  more specific wording now says that resent request must have the same
  sequence number as the original.

  Also see Page 39, S12.17

Page 28, S10.6

  Added to the end of first paragraph: 
    "Any server resources are kept, though servers MAY close the session
    and free resources after being paused for the duration specified with
    the {\sf timeout} parameter of the {\sf Session} header in the {\sf
    SETUP} message.  

  Example moved to after first paragraph to make things slightly clearer.

Page 30, S10.9
  Last paragraph of text moved to after the example to avoid confusion.

  Added Content-Type field to the response, and added Content-Length field
  to the request.

Page 33, S11.3.6

  Added the following sentence: "The response SHOULD contain an Allow
  header to make error recovery easier."

Page 34, S11.3.14
  State that an "Unsupported" field is returned.

Page 34, S12
  Third paragraph removed:

  "If the field content does not apply to the particular resource, the
  server MUST return status 456 (Header Field Not Valid for Resource)."

Page 34 S12 para 2
  "Note that not all fields marked "r" ..."  
   .. here and in the next sentence "r" changed to "req".

Page 37  BNF
   In "cache-extension", "---" changed to "|" (.ps only)

Page 41, S12.29
  "This specification defines..." now includes NPT

  Added "If the Range header is given in a time format that is not
  understood, the recipient should return ``501 Not Implemented''."

Page 42, S12.32
  Added note explaining why "funky-feature" and "funky-parameter" are
  not the same.

Page 43, S12.33
  Added rtptime to the BNF

Page 45, S12.39 
  "A server may refuse to change parameters of an existing stream".
  What error code does it return to do this?

Page 45, S12.39 "source"
  Why is source only valid for unicast?  It seems this might be valid
  for multicast too as a small degree of protection against spoof
  sources.

Page 46, S12.39 "interleaved"
  "eg."->"e.g.", other grammatical tweaks.

Page 47, S12.39
  The BNF should allow multiple comma separated transport specs but doesn't.

Page 49, S14.1
  in the SDP, the video URL should be video.example.com not audio.example.com

Page 50
  "CSeq: 4" in the last request and response should be "CSeq: 3"

Page 53, S14.3
   Why does CSeq start with 2?

Page 54, example at top of page
   The indenting is messed up

Page 55, setup response
   The port for RTP/RTCP should be a range 3456-3457

Page 56, S14.6
   The conference participant is recording both audio and video, not just
   "the audio portion".

Page 57, video setup request and reply
   The multicast address for video should be 224.0.1.12 not .11

Page 57, RECORD request
   The Range field is missing an "=" after the word "clock".

Page 58, S15.1
   Messed up formatting in "OCTET", "quoted-string", "qdtext", and
   "quoted-pair"