[XeTeX] fontspec manual 10.17 (scripts/languages)

[François Charette]

  On 19/07/2010 21:10, David Perry wrote:
> I've been reading 10.17 (version of July 11, 2010) and have some questions, and found 
> one typo.

I will let Will comment on your interesting feedback. I only have a few minor observations 
below.

>
> 10.17 (the very beginning)
> It would be helpful to have an example of the script and language selection in use.  I 
> use the Junicode font, which contains an OT lookup to provide the Icelandic form of the 
> letter thorn rather than the Old English form when text is tagged as Icelandic.  What is 
> wrong with this?
>
>     \fontspec[Language=Icelandic]{Junicode}
>     Þis is the letter þorn in the Icelandic language, using Junicode
>
> It compiles without protest, but I still get the default Old English shape for thorn.  
> (My document is already using Latin script, so I didn't specify that specifically.)
>
> There may also be some interactions of script and language.  A friend asked me to help 
> him get Sanskrit working in XeLaTeX. He's using the Devanagari script (I am told that 
> Sanskrit is sometimes written in other scripts). He's on a Mac using an AAT font, so he 
> should only need to use:
>
> \fontspec{Devanagari MT}
>
> right?  At first I told him to include
>
> \fontspec[Script=Devanagari]{Devanagari MT}
>
> but later realized that's only for OT.  The second paragraph of 10.17 does mention this, 
> but a footnote or parenthetical remark ("Don't do this with AAT fonts!") might save 
> people like me who work almost exclusively with OT from making silly mistakes.

Hmm... I realize that polyglossia implicitly assumes OT fonts only: I should add a warning 
when encountering AAT or Graphite fonts. The new fontspec has convenient macros to check 
for that I think.

>
> I take it he does NOT need to specify [Language=Sanskrit].  Later on, if he gets an OT 
> font that has some special features for Sanskrit (if such a font exists--I'll call it 
> RigVeda) and wants to use it on his Mac, then he would write:
>
> \fontspec[Script=Devanagari,Language=Sanskrit]{RigVeda}
>
> to automatically select the ICU renderer and enable the special stuff for Sanskrit.  Yes?
>
> 10.17.1
> Typo: TRK is the tag for Turkish in OpenType; tur is the ISO tag.  (See 
> http://www.microsoft.com/typography/otspec/ttoreg.htm) 

Yet many "real-world" OT fonts mistakenly use TUR instead of TRK for Turkish, hence in 
this particular case fontspec needs to check for both when "Language=Turkish" is specified.

> A brief explanation beyond the phrase "OpenType definition" might be helpful too for 
> those who don't know the OT tags yet (especially since they are not the same as the ISO 
> tags with which people might be familiar).
>
> The phrase "Further scripts . . . " confused me.  First I thought it meant the ability 
> to add scripts or languages not already defined in fontspec.  But fontspec already knows 
> all the ones that are defined in OpenType, so how could one add a new script if it 
> doesn't have an OT tag, since the OT tag is required for the definition?  And the 
> examples showed a script and a language that are already in fontspec.  Does it mean 
> something different?
>

I think (Will can correct me if I'm wrong) these are generic macros used internally. They 
are nevertheless accessible to the end-user, just in case some font (designed, say, for 
Klingon or, perhaps more seriously, for specialized ethnolinguistic uses) were to use 
language and/or script tags not defined in the current OpenType standard. A sentence 
clarifiying this might be helpful though.

FC