[OpenType] [mpeg-OTspec] RE: Long-standing typo in 'OS/2' table section

[Ken Lunde]

Hin-Tak,

The correct range is for Mahjong Tiles is 1F000-1F02F, which is reflected in the online OpenType Specification and in ISO/IEC 14496-22:2015. You can also confirm this on the SMP's roadmap page:

  http://www.unicode.org/roadmaps/smp/

With regard to Plane 16, the last two code points in the BMP and in all 16 additional planes, meaning FFFE and FFFF in the BMP and *FFFE and *FFFF in the additional planes, are noncharacters.

Related to this I also noted the following in both the online specification and in the ISO standard:

69 Specials FFF0-FFFF

FFFF should be FFFD if noncharacters are excluded. Also consider that 32 noncharacters in the range UFDD0 through U+FDEF are included in the entry below, which is also reflected in both the online specification and in the ISO standard:

63 Arabic Presentation Forms-A FB50-FDFF

But, for the purpose of the 'OS/2' table specification, there is no harm in including noncharacters, and doing so also looks cleaner. So, 90 would become the following (changing FFFFD and 10FFFD to FFFFF and 10FFFF, respectively:

90 Private Use (plane 15) F0000-FFFFF
   Private Use (plane 16) 100000-10FFFF

And, 63 and 69 are therefore okay as-is.

Best...

-- Ken

> On Nov 16, 2016, at 4:02 AM, Hin-Tak Leung <htl10 at users.sourceforge.net> wrote:
> 
> Message from OpenType list:
> 
> 
> Argh, ouch.
> 
> People at Microsoft (or at least the person who wrote the FontVal code)
> knew the correct version, regardless of the typo in the spec:
> 
> https://github.com/Microsoft/Font-Validator/tree/master/OTFontFileVal/val_OS2.cs#L1266
> 
> When I updated the code for OS/2 v4, I sync'ed against the current spec, and broke it:
> 
> https://github.com/HinTak/Font-Validator/master/OTFontFileVal/val_OS2.cs#L1329
> 
> There is one other big difference when I syn'ed the table - the 'Mahjong Tiles' range.
> 
> where should it start, 0x1f020 or 0x1f000 ?
> 
> I assumed the code was wrong as it was for os2 v3, the spec of which wasn't too precise about the ranges.
> (and a small difference about plane 16 ending on 0x10ffff vs 0x10fffd - this I am more certain that the old code was probably wrong).
> 
> In practice, those code differences probably aren't important, since the ranges are only used for glyph coverage.
> 
> --------------------------------------------
> On Tue, 15/11/16, Peter Constable <petercon at microsoft.com> wrote:
> 
> Message from OpenType list:
> 
> 
> The OT 1.8 spec pages for OS/2 and OS/2 v4 have been
> updated. This will also be listed on the errata page (still
> in process).
> 
> 
> Peter
> 
> 
> 
> List archive: http://www.indx.co.uk/biglistarchive/
> List settings: http://www.indx.co.uk/biglistarchive/?mode=usersettings
> 
> subscribe: opentype-subscribe at indx.co.uk
> unsubscribe: opentype-unsubscribe at indx.co.uk
> messages: opentype-list at indx.co.uk
> 
>