Meeting Notes 20070727

Meeting at Unicode/Microsoft office, Mountain View, 2007-Jul-27

Participants:

Kat Momoi (Google)

Markus Scherer (Google)

Mark Davis (Google)

Ken Whistler (Sybase)

Present:

Rick McGowan (Unicode)

Eric Muller (Adobe)

Joe Becker (Unicode)

Criteria for list of Emoji symbols

- Source separation rules

- Follow existing cross-mapping tables

+ but tentatively disunified in some cases where the visual images were very different,

e.g., M symbol for Metro vs. Metro train image

- List selects any one of the multiple carrier's images, but for Unicode encoding

representative glyphs would have to be created

- Roundtrip (1:1) mapped symbols among carriers are unmarked (no background color)

- Best-fit fallback mappings are marked with gold background and a *

- Fallback mappings to sequences of codes are marked with + between the codes

- Fallback mappings to descriptive text rather than a symbol is marked with purple background

- Unicode character names: "black" and "white" usually mean "solid fill" vs. "hollow fill";

therefore, only mention color in a symbol name if the source-separation rule requires a distinction

Ken:

- Principle 1: Encoding symbols as characters abstracts away from colors.

Only use colors for distinction for the source-separation rule.

- Principle 2: Abstract away from animated behavior.

Only use animation for distinction for the source-separation rule.

Animated vs. un-animated hourglass:

- Ken: Map existing Unicode code point to the set of symbols that roundtrip between carriers,

and assign a new one for the symbols which use fallbacks.

Map to "least marked alternative".

Put "animation" into comment, not into character name.

- Mark would map existing Unicode code point to un-animated symbol, but not a strong opinion.

Consensus to follow Ken's suggestion.

Animation and other distinctions in general:

- Mark: If the distinction is relatively clear from the symbols, put it into the names,

otherwise number them (or similar) and put the distinctions into the comments.

Double-check names of existing Unicode characters (e.g., watch, full moon)

Moon phases:

- Ken: Carriers do not distinguish crescent moon phases, but Unicode does.

Therefore, encode a new symbol for "crescent moon" and comment that the orientation

does not matter.

Zodiac symbols:

- Ken: Unify animals with astrological symbols

- Mark: As a result, Unicode Zodiac symbols are allowed to be displayed with animal images

- Ken: Zodiac symbols, and characters in general, have a substantial range of glyph variation

- Mark: Want to discuss this in a larger group because of its implications

- Ken: As a linguist, this is a no-brainer

- Mark: Disunification in Unicode does not prevent roundtrips between carriers.

Mappings are possible like

x<->u1 ->y

x<- u2<->y

while carriers map

x<->y

- Ken: Avoid creating whole new signification for encoding distinctions.

- Mark: Can live with unification, especially for *sets of things*.

Minimize disunification for sets of corresponding entities.

Kat: Right now, hard to get buy-in from Japanese mobile carriers to change anything.

Mark: We respect the existing mappings between the carriers.

Eric: Unicode may be used on phones, which needs mapping tables.

Enclosed M vs. Metro:

- Ken: Enclosed M presents translation problem: Would have to be enclosed U in German.

- Mark: Safer to encode enclosed M and enclosed U and subway train separately,

like for currency symbols.