Adding RAINBOW FLAG to Unicode

Philippe Verdy verdy_p at
Sat Jun 27 17:40:32 CDT 2015

No I had read it, the persons are still clearly separate. The "rainbow" on
the flag is not in fact a rainbow, only its colours.

The groups of persons are showing persons themselves, side by side, not one
into the other one or one indirectly drawn on the face of another one.

This new proposal of use of ZWJ is *definitely NOT in use*, it assumes a
strong alteration of semantics. What is represented is NOT a flag and a
rainbow side by side. The physical natural phenomenum (and its real 3D cone
shape) is NOT represented at all on the flag, but the flag also adds
parallel stripes (not encoded by the rainbow symbols itself and not by
the white flag symbol alone: if you need to use a country flag to have
these bands, you'll add a country-specific semantic that is not part of the
international flag).

The case is different from the black flag wih skulls and cross bones: what
is represented is effectively a realistic skull and cross bones, not just
the color or impression left by these bones.

The nearest equivalent you can see is with the fitzpatrick modifers for
skin colours (that do NOT use any ZWJ, because the fitzepatrick are
intended to be modifiers and have no shape semantics by themselves).

The only interpretation of FLAG + ZWJ + RAINBOW can only be two separate
objects side by side (or one partly covering the other one) (like in the
Family examples).

(there's already a trong resistance for just embedding letters on a flag,
country flags had then to be encoded differently, and letters enclosed in
other shapes such as boxes (similar to the flag) are using combining boxes:
we would need a combining flag character to do that (but before this
happens we need a way to create "cartouches" for hieroglyphs or sinograms
or even latin letters: this doid not occur, and instead emojis are encoding
these enclosed letters distinctly without using any sequence (with
combining characters or with joiners). For the same reason overstriking
combining characters are best avoided for letters (this causes
interpretation problems).

You can expect interpretation problems if you intend to use ZWJ to create a
ligature that completely drops the esssential shape of the rainbow to keep
only its colors in a tiny part of it. by evidence this flag is NOT a

Or otherwise, country flags are ALL ligatures (even if they don't represent
the two letters with which they were internally encoded, they don't contain
these letters and don't have the semantics of these letters, all that is
meant is an assopciation with a country name, and then with its current

If we only wanted to include the semantics of the colour, then we would not
even need fitzpatrick modifiers, we would have used ZWJ with white or black
filled shapes (boxes, discs, independantly of their size and shapes...).
ZWJ is NOT a semantics killer.

2015-06-27 23:48 GMT+02:00 Peter Edberg <pedberg at>:

> Philippe and others,
> You are missing the relevant parts of UTR #51. See:
> This type of behavior with ZWJ for emoji is *already in use.*
> - Peter E
> On Jun 27, 2015, at 1:14 PM, Philippe Verdy <verdy_p at> wrote:
> This UTR just addresses the case of a combining coloring symbol for faces
> and those color symbols were designed since the begining to be combined as
> much as possible (and not meant to be used in isolation), this is not the
> case of the rainbow symbol which is much more figurative).
> Why would associating a flag and a rainbow this way means the flag will
> just be recolored (but the rainbox form itself is completely lost)?
> Couldn't this be to display a flying flag over a sky with a rainbow?
> Compare this to the association of the sun and the rainbow symbols, or the
> cloud and a rainbow (and compare to the sun or moon and a cloud associated
> the same way, or the association of two clouds: none of them will overlap
> completely).
> Imagine the use in a weather application, I don't wee why the rainbox
> would disappear when the flying flag is just there to mean the windy
> condition, and the rainbox meant for variable weather mixing rainy and
> sunny periods.
> Your proposed use of ZWJ to create a complete overlap of one symbol into
> another is unexpected.
> ZWJ+symbol does not transfor that symbol into a "emoi modifier" (this is
> not anywhere in UTF51). It may just create a small partial overlap of one
> symbol into the other, but each one is still clearly identifiable
> separately. The examples shown are for grouping multiple persons in Annex E
> but each person is still separately visible and recognizable as such even
> if they are combined in the same final glyph. Annexe E even requires some
> specific orders (e.g. for families: the man can only come before a woman,
> and is then necessarily visible to the left side of the icon, i.e. to the
> right of the woman; children are necessarily after and below adults...).
> 2015-06-27 21:31 GMT+02:00 Mark Davis [image: ☕]️ <mark at>:
>> Take a look at for details.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: emoji_u2615.png
Type: image/png
Size: 2776 bytes
Desc: not available
URL: <>

More information about the Unicode mailing list