Hi Richard, On 16/10/18 2:29 pm, Richard Wordingham via Unicode wrote: > On Tue, 16 Oct 2018 11:59:54 +1100 > Harshula via Unicode <[email protected]> wrote: > >> Hi Richard, >> >> On 16/10/18 6:57 am, Richard Wordingham via Unicode wrote: >>> On Tue, 16 Oct 2018 02:47:36 +1100 >>> Harshula via Unicode <[email protected]> wrote: >>> >>>> Note, touching letters are formed by <ZWJ><AL-LAKUNA>, so they >>>> should not be displayed as a fallback for <AL-LAKUNA><ZWJ> >>>> conjuncts. >>> >>> I don't follow that. While the conjuncts with r-, -r and -y are >>> very different to pairs of touching letters, the conjuncts for tth, >>> nd, ndr, ndh, kv and tv would be very similar to the hypothetical >>> corresponding touching letters and quite different to the fallbacks >>> with visible al-lakuna. >> >> If you haven't already, it's best you read SLS 1134:2011: >> http://www.language.lk/en/download/standards/ >> >> or the older SLS 1134:2004: >> http://unicode.org/wg2/docs/n2737.pdf > > The latter actually says, in Section 5.8, that <AL-LAKUNA, ZWJ> may be > used for either! I suspect that that is a printing error.
The former (SLS1134:2011) has a section for Touching letters. It is explicitly stated to use <ZWJ><AL-LAKUNA> for Touching letters. Sorry, the file n2737.pdf hosted on unicode.org appears to be a draft. It is not the final SLS1134:2004. The final contains a section on Touching letters like SLS1134:2011. > The Sri Lankan standard simply assumes that the rendering system can > accommodate what is requested in the backing store. It says nothing > about fallbacks. So, if the user specifies the the syllable ddho > written with a conjunct and encoded as ද්ධො but the conjunct is > missing from the fonts' repertoires, why is it right to display it with > al-lakuna as though it were ද්ධො but wrong to display it with the > touching letters encoded as ද්ධො? There are three different > correct ways of writing 'ddho', but many systems only support one of > them (and some weirdly use a fourth method). When a font is missing a glyph during an *explicit* conjunct lookup, it appears the most accurate solution is to display the missing glyph symbol. cya, #
