ISSUE-259: Define "closest"

Closest

Define "closest"

State:
CLOSED
Product:
TTML2
Raised by:
Pierre-Anthony Lemieux
Opened on:
2013-06-27
Description:
The term "closest" is used a number of times. Does closest mean 12.5 = 12 or 12.5 = 13? Suggest defining closest more precisely.
Related Actions Items:
Related emails:
  1. {minutes} TTWG Meeting 24/7/2014 (from nigel.megitt@bbc.co.uk on 2014-07-24)
  2. {agenda} TTWG Meeting 24/7/2014 (from nigel.megitt@bbc.co.uk on 2014-07-23)
  3. Re: Need Update to TTML1 Errata (from nigel.megitt@bbc.co.uk on 2014-07-21)
  4. Re: Need Update to TTML1 Errata (from pal@sandflow.com on 2014-07-20)
  5. Re: Need Update to TTML1 Errata (from glenn@skynav.com on 2014-07-17)
  6. Re: Need Update to TTML1 Errata (from nigel.megitt@bbc.co.uk on 2014-07-17)
  7. {minutes} TTWG Meeting 17/7/2014 (from nigel.megitt@bbc.co.uk on 2014-07-17)
  8. {agenda} TTWG Meeting 17/7/2014 (from nigel.megitt@bbc.co.uk on 2014-07-16)
  9. {agenda} TTWG Meeting 10/7/2014 (from nigel.megitt@bbc.co.uk on 2014-07-09)
  10. RE: {agenda} TTWG Meeting 19/6/2014 (from mdolan@newtbt.com on 2014-06-18)
  11. {agenda} TTWG Meeting 19/6/2014 (from nigel.megitt@bbc.co.uk on 2014-06-18)
  12. RE: {agenda} TTWG Meeting 12/6/2014 (from mdolan@newtbt.com on 2014-06-12)
  13. Re: {agenda} TTWG Meeting 12/6/2014 (from pal@sandflow.com on 2014-06-11)
  14. Re: {agenda} TTWG Meeting 12/6/2014 (from nigel.megitt@bbc.co.uk on 2014-06-11)
  15. Re: {agenda} TTWG Meeting 12/6/2014 (from pal@sandflow.com on 2014-06-11)
  16. Re: {agenda} TTWG Meeting 12/6/2014 (from nigel.megitt@bbc.co.uk on 2014-06-11)
  17. {agenda} TTWG Meeting 12/6/2014 (from nigel.megitt@bbc.co.uk on 2014-06-11)
  18. RE: {agenda} TTWG Meeting 5/6/2014 (from mdolan@newtbt.com on 2014-06-04)
  19. {agenda} TTWG Meeting 5/6/2014 (from nigel.megitt@bbc.co.uk on 2014-06-04)
  20. {agenda} TTWG Meeting 29/5/2014 (from nigel.megitt@bbc.co.uk on 2014-05-28)
  21. {agenda} TTWG Meeting 22/5/2014 (from nigel.megitt@bbc.co.uk on 2014-05-21)
  22. {agenda} TTWG Meeting 15/5/2014 (from nigel.megitt@bbc.co.uk on 2014-05-14)
  23. Re: {agenda} TTWG Meeting 8/5/2014 (from tmichel@w3.org on 2014-05-08)
  24. {agenda} TTWG Meeting 8/5/2014 (from nigel.megitt@bbc.co.uk on 2014-05-07)
  25. {agenda} TTWG Meeting 1/5/2014 (from nigel.megitt@bbc.co.uk on 2014-04-30)
  26. {agenda} TTWG Meeting 24/4/2014 (from nigel.megitt@bbc.co.uk on 2014-04-23)
  27. {agenda} TTWG Meeting 17/4/2014 (from nigel.megitt@bbc.co.uk on 2014-04-17)
  28. Re: {agenda} TTWG Meeting 27/3/2014 (from glenn@skynav.com on 2014-03-27)
  29. Re: {agenda} TTWG Meeting 27/3/2014 (from tmichel@w3.org on 2014-03-27)
  30. {agenda} TTWG Meeting 27/3/2014 (from nigel.megitt@bbc.co.uk on 2014-03-26)
  31. {minutes} TTWG Meeting 27/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-27)
  32. ISSUE-300 (tts:origin correct fallback): Prior interpretation of the correct fallback for tts:origin [TTML2] (from sysbot+tracker@w3.org on 2014-02-26)
  33. {agenda} TTWG Meeting 27/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-26)
  34. {minutes} TTWG Meeting 20/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-20)
  35. {agenda} TTWG Meeting 20/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-19)
  36. Re: {minutes} TTWG Meeting 13/2/2014 (from tmichel@w3.org on 2014-02-13)
  37. {minutes} TTWG Meeting 13/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-13)
  38. RE: {agenda} TTWG Meeting 13/2/2014 (from mdolan@newtbt.com on 2014-02-13)
  39. Re: {agenda} TTWG Meeting 13/2/2014 (from glenn@skynav.com on 2014-02-13)
  40. {agenda} TTWG Meeting 13/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-12)
  41. Re: {agenda} TTWG Meeting 6/2/2014 (from silviapfeiffer1@gmail.com on 2014-02-06)
  42. Re: {agenda} TTWG Meeting 6/2/2014 (from silviapfeiffer1@gmail.com on 2014-02-06)
  43. Re: {agenda} TTWG Meeting 6/2/2014 (from singer@apple.com on 2014-02-05)
  44. {agenda} TTWG Meeting 6/2/2014 (from nigel.megitt@bbc.co.uk on 2014-02-05)
  45. {minutes} TTWG Meeting 30/01/2014 (from nigel.megitt@bbc.co.uk on 2014-01-30)
  46. {agenda} TTWG Meeting 30/01/2014 (from nigel.megitt@bbc.co.uk on 2014-01-30)
  47. {minutes} 16/1/14 TTWG meeting (from nigel.megitt@bbc.co.uk on 2014-01-16)
  48. {agenda} 16/1/14 TTWG meeting (from nigel.megitt@bbc.co.uk on 2014-01-16)
  49. {minutes} 9/1/14 TTWG Meeting (from nigel.megitt@bbc.co.uk on 2014-01-09)
  50. TTWG Agenda for 9/1/14 (from nigel.megitt@bbc.co.uk on 2014-01-08)
  51. Minutes for 12/12/13 (from nigel.megitt@bbc.co.uk on 2013-12-12)
  52. TTWG Agenda for 12/12/13 (from nigel.megitt@bbc.co.uk on 2013-12-11)
  53. Re: TTWG Agenda for 5/12/13 (from nigel.megitt@bbc.co.uk on 2013-12-11)
  54. Re: TTWG Agenda for 5/12/13 (from glenn@skynav.com on 2013-12-11)
  55. RE: TTWG Agenda for 5/12/13 (from mdolan@newtbt.com on 2013-12-10)
  56. TTWG Agenda for 5/12/13 (from nigel.megitt@bbc.co.uk on 2013-12-04)
  57. Re: TTML Agenda for 21/11/13 (from cyril.concolato@telecom-paristech.fr on 2013-11-21)
  58. Re: TTML Agenda for 21/11/13 (from glenn@skynav.com on 2013-11-21)
  59. Re: TTML Agenda for 21/11/13 (from tmichel@w3.org on 2013-11-21)
  60. TTML Agenda for 21/11/13 (from nigel.megitt@bbc.co.uk on 2013-11-20)
  61. RE: TTML Agenda for 31/10/13 (from Sean.Hayes@microsoft.com on 2013-10-31)
  62. TTML Agenda for 31/10/13 (from nigel.megitt@bbc.co.uk on 2013-10-30)
  63. {minutes} TTML Meeting of 24/10/13 (from glenn@skynav.com on 2013-10-24)
  64. RE: TTML Agenda for 24/10/13 (from mdolan@newtbt.com on 2013-10-23)
  65. TTML Agenda for 24/10/13 (from nigel.megitt@bbc.co.uk on 2013-10-23)
  66. {minutes} TTML Meeting of 17/10/13 (from glenn@skynav.com on 2013-10-17)
  67. TTML Agenda for 17/10/13 (from Sean.Hayes@microsoft.com on 2013-10-17)
  68. TTML Agenda for 03/10/13 (from Sean.Hayes@microsoft.com on 2013-10-03)
  69. Re: TTML Minutes for 25/09/13 (from tmichel@w3.org on 2013-09-27)
  70. RE: TTML Minutes for 25/09/13 (from silviapfeiffer1@gmail.com on 2013-09-27)
  71. RE: TTML Minutes for 25/09/13 (from nigel.megitt@bbc.co.uk on 2013-09-26)
  72. TTML Agenda for 25/09/13 (from Sean.Hayes@microsoft.com on 2013-09-25)
  73. TTML Agenda for 19/09/13 (from Sean.Hayes@microsoft.com on 2013-09-19)
  74. No Meeting today 12-09-13 (from Sean.Hayes@microsoft.com on 2013-09-12)
  75. Minutes for 05/09/13 (from nigel.megitt@bbc.co.uk on 2013-09-05)
  76. Agenda for 05/09/13 (from glenn@skynav.com on 2013-09-04)
  77. RE: TTML Agenda for 29/08/13 (from mdolan@newtbt.com on 2013-08-28)
  78. TTML Agenda for 29/08/13 (from glenn@skynav.com on 2013-08-28)
  79. TTML Agenda for 15/08/13 (from Sean.Hayes@microsoft.com on 2013-08-15)
  80. Re: TML Agenda for 06/08/13 (from glenn@skynav.com on 2013-08-08)
  81. [Minutes] TTWG telecon 08/08/2013 (from tmichel@w3.org on 2013-08-08)
  82. Re: TML Agenda for 06/08/13 (from pal@palemieux.com on 2013-08-08)
  83. TML Agenda for 06/08/13 (from Sean.Hayes@microsoft.com on 2013-08-08)
  84. [minutes] Timed Text 20130725 (from plh@w3.org on 2013-07-25)
  85. TTML Agenda for 25/07/13 (from Sean.Hayes@microsoft.com on 2013-07-25)
  86. ISSUE-259 (Closest): Define "closest" [TTML.next] (from sysbot+tracker@w3.org on 2013-06-27)

Related notes:

Leave open until change effected.

Glenn Adams, 20 Aug 2013, 15:56:49

Fixed. https://dvcs.w3.org/hg/ttml/rev/24ed1e6adcdd

Glenn Adams, 28 Aug 2013, 18:21:16

[glenn]: need to add "round-down" clarification, then close

19 Sep 2013, 15:37:20

Suggest using 1SE definitions as basis.

For scalar values, suggest:

"then a presentation processor must use the closest supported value unless computed value is equidistant from two supported values, in which case the processor must use the one of the two supported valued that is closest to +infinity, i.e. round half up"

For a vector value, e.g. extent, suggest treating each vector coordinate separately as a scalar per the above.

For <color>, suggest

(i) treating alpha as a scalar, separately from rgb
(ii) for <rgb>, "then a presentation processor must use the nearest supported value of <rgb> in the euclidian (r,g,b) space unless the computed value is equidistant from two supported values, in which case the processor must use the one of the two supported values farthest from (0,0,0) in the euclidian (r,g,b) space."

Pierre-Anthony Lemieux, 24 Oct 2013, 14:13:03

[glenn]: address new input from pal

24 Oct 2013, 14:13:20

PAL: P2

Philippe Le Hégaret, 24 Oct 2013, 15:59:15

Both TTML1 and TTML2 already deal with issue of closest value when multiple nearest values are present. No further action is required.

Glenn Adams, 5 Dec 2013, 20:43:36

Is there any reason not to use round half up consistently, e.g. in tts:origin, the following definition is used:

"closest supported value means the value for which the Euclidean distance between the computed origin and the supported origin is minimized."

I also suggest that, for vector quantities, applying round half up to each quantity might be easier to understand for implementers than "closest euclidian distance", square root and al.

Pierre-Anthony Lemieux, 12 Dec 2013, 16:13:06

[nigel]: Glenn: it doesn't always make sense to round up, sometimes round down works better, e.g. zIndex

9 Jan 2014, 16:24:22

[nigel]: pal: I will include some specific text proposals.

9 Jan 2014, 16:25:30

Specific suggestions.

Option (a) [minimal]

For tts:origin, specify "if there are multiple closest supported values equally distant from the computed value, then the value most distant from [0,0], i.e. most distant to the coordinate space origin, is used."

This matches other length-based properties, e.g. extent, padding, lineHeight...

Option (b) [suggested]

For tts:origin, tts:extent, tts:fontSize and tts:padding, specify

"""
If a computed <length> value(s) is not supported, then a presentation processor must use the closest supported value.

Note:

In this context, the phrase closest supported value means the value for which the absolute distance between the computed <length> value and the supported <length> value is minimized. If there are multiple closest supported values equally distant from the computed value, then the supported value most distant from 0 is used.
"""

By considering each <length> value separately, complex euclidean distance computations are avoided.

Pierre-Anthony Lemieux, 16 Jan 2014, 06:52:53

[nigel]: Scalar and vector comparisons need to be distinguished more clearly.

30 Jan 2014, 15:40:44

Nigel: need to be explicit about whether comparisons should be done vector-wise vs per-scalar value (e.g. the calculation for fontSize is vector, whereas padding is a set of up to 4 scalars)

Pierre-Anthony Lemieux, 26 Feb 2014, 22:21:43

Clarify scalar interpretation for border, padding. Improve language for extent. Add errata for TTML1.

https://dvcs.w3.org/hg/ttml/rev/52998d4ca12c

Glenn Adams, 17 Jul 2014, 04:20:16

Display change log ATOM feed

David Singer <singer@apple.com>, Nigel Megitt <nigel.megitt@bbc.co.uk>, Chairs, Thierry Michel <tmichel@w3.org>, Philippe Le Hégaret <plh@w3.org>, Atsushi Shimono <atsushi@w3.org>, Staff Contacts
Tracker: documentation, (configuration for this group), originally developed by Dean Jackson, is developed and maintained by the Systems Team <w3t-sys@w3.org>.
$Id: 259.html,v 1.1 2019/11/12 10:06:31 carcone Exp $