moyogo
commented
10 months ago @EbenSorkin The accented letters that are not available as single characters are the following, based on gflanguages data :
A̋ A̧ A̰ A̱ A᷆ A᷇ B̀ B́ C̱ E̋ E̱ E᷆ E᷇ G̈ I̋ I̧ I̱ I᷆ I᷇ J̌ L̀ M̀ M̄ N̂ N̄ N̈ N̰ O̧ O̱ O᷆ O᷇ P̱ R̀ R̂ S̀ T͟H U̧ U̱ U᷆ U᷇ W̤ a̋ a̧ a̰ a̱ a᷆ a᷇ b̀ b́ c̱ e̋ e̱ e᷆ e᷇ g̈ i̋ i̧ i̱ i᷆ i᷇ l̀ m̀ m̄ n̂ n̄ n̈ n̰ o̧ o̱ o᷆ o᷇ p̱ r̀ r̂ s̀ t͟h u̧ u̱ u᷆ u᷇ w̤ À̰ Á̧ Á̰ Â̧ Ã̀ Ã́ Ã̌ Æ̀ Æ̂ Æ̋ Æ̌ Æ̏ Í̧ Î̧ Ó̧ Ô̧ Õ̀ Õ̌ Õ̱ Ø̂ Ø̌ Ú̧ Û̧ à̰ á̧ á̰ â̧ ã̀ ã́ ã̌ æ̀ æ̂ æ̋ æ̌ æ̏ í̧ î̧ ó̧ ô̧ õ̀ õ̌ õ̱ ø̂ ø̌ ú̧ û̧ Ĩ̀ Ĩ́ Ĩ̌ ĩ̀ ĩ́ ĩ̌ Ŋ̀ Ŋ́ Ŋ̄ Ŋ̈ ŋ̀ ŋ́ ŋ̄ ŋ̈ Œ̀ Œ́ Œ̂ Œ̄ Œ̋ Œ̌ Œ̏ œ̀ œ́ œ̂ œ̄ œ̋ œ̌ œ̏ Ũ̀ Ũ̌ ũ̀ ũ̌ Ɔ̀ Ɔ́ Ɔ̂ Ɔ̃ Ɔ̃̀ Ɔ̃́ Ɔ̃̌ Ɔ̄ Ɔ̈ Ɔ̌ Ɔ̧ Ɔ̧́ Ɔ̧̂ Ɔ̰ Ɔ̰̀ Ɔ̰́ Ɔ̱ Ɔ᷆ Ɔ᷇ Ǝ̀ Ǝ́ Ǝ̂ Ǝ̃ Ǝ̃̀ Ǝ̃́ Ǝ̄ Ǝ̌ Ə̀ Ə́ Ə̂ Ə̄ Ə̈ Ə̌ Ɛ̀ Ɛ́ Ɛ̂ Ɛ̃ Ɛ̃̀ Ɛ̃́ Ɛ̃̌ Ɛ̄ Ɛ̈ Ɛ̋ Ɛ̌ Ɛ̏ Ɛ̧ Ɛ̧́ Ɛ̧̂ Ɛ̰ Ɛ̰̀ Ɛ̰́ Ɛ̱ Ɛ̱̈ Ɛ᷆ Ɛ᷇ Ɩ̀ Ɩ́ Ɩ̃ Ɩ̃̀ Ɩ̃́ Ɨ̀ Ɨ́ Ɨ̂ Ɨ̄ Ɨ̌ Ɯ̀ Ɯ́ Ɯ̂ Ɯ̄ Ɯ̋ Ɯ̏ Ɵ̀ Ɵ́ Ɵ̂ Ʋ̀ Ʋ́ Ʋ̃ Ʋ̃̀ Ʋ̃́ ǝ̀ ǝ́ ǝ̂ ǝ̃ ǝ̃̀ ǝ̃́ ǝ̄ ǝ̌ Ȩ́ Ȩ̂ ȩ́ ȩ̂ Ʉ̀ Ʉ́ Ʉ̂ Ʉ̄ Ʉ̈ Ʉ̌ Ʌ̀ Ʌ́ Ʌ̂ Ʌ̄ Ʌ̋ Ʌ̏ ɑ̀ ɑ́ ɑ̂ ɑ̄ ɑ̌ ɔ̀ ɔ́ ɔ̂ ɔ̃ ɔ̃̀ ɔ̃́ ɔ̃̌ ɔ̄ ɔ̈ ɔ̌ ɔ̧ ɔ̧́ ɔ̧̂ ɔ̰ ɔ̰̀ ɔ̰́ ɔ̱ ɔ᷆ ɔ᷇ ə̀ ə́ ə̂ ə̄ ə̈ ə̌ ɛ̀ ɛ́ ɛ̂ ɛ̃ ɛ̃̀ ɛ̃́ ɛ̃̌ ɛ̄ ɛ̈ ɛ̋ ɛ̌ ɛ̏ ɛ̧ ɛ̧́ ɛ̧̂ ɛ̰ ɛ̰̀ ɛ̰́ ɛ̱ ɛ̱̈ ɛ᷆ ɛ᷇ ɤ̀ ɤ́ ɤ̂ ɤ̄ ɤ̋ ɤ̏ ɥ̀ ɥ́ ɥ̂ ɨ̀ ɨ́ ɨ̂ ɨ̄ ɨ̌ ɩ̀ ɩ́ ɩ̃ ɩ̃̀ ɩ̃́ ɪ̃ ɯ̀ ɯ́ ɯ̂ ɯ̄ ɯ̋ ɯ̏ ɵ̀ ɵ́ ɵ̂ ʉ̀ ʉ́ ʉ̂ ʉ̄ ʉ̈ ʉ̌ ʋ̀ ʋ́ ʋ̃ ʋ̃̀ ʋ̃́ ʌ̀ ʌ́ ʌ̂ ʌ̄ ʌ̋ ʌ̏ Ḭ̀ Ḭ́ ḭ̀ ḭ́ Ṵ̀ Ṵ́ ṵ̀ ṵ́ Ạ́ Ạ̃ ạ́ ạ̃ Ẹ̀ Ẹ́ Ẹ̃ Ẹ̄ ẹ̀ ẹ́ ẹ̃ ẹ̄ Ẽ̀ Ẽ́ Ẽ̌ Ẽ̱ ẽ̀ ẽ́ ẽ̌ ẽ̱ Ị̀ Ị́ Ị̂ Ị̃ Ị̄ ị̀ ị́ ị̂ ị̃ ị̄ Ọ̀ Ọ́ Ọ̃ Ọ̄ ọ̀ ọ́ ọ̃ ọ̄ Ụ̀ Ụ́ Ụ̂ Ụ̃ Ụ̄ ụ̀ ụ́ ụ̂ ụ̃ ụ̄ Ɑ̀ Ɑ́ Ɑ̂ Ɑ̄ Ɑ̌ Ɥ̀ Ɥ́ Ɥ̂ Ɪ̃ Ꞷ̃ ꞷ̃
EbenSorkin
I ( Eben) asked Denis : I have been updating the samples for the Africa Pri projects in Octotext, but it occurred to me to ask you Denis:
• Are there languages which we should now be sure to include paragraphs for texture reasons since they have been included in Africa Priority even if they are maybe not so different or helpful in terms of displaying use of particular african specific glyphs? The paragraphs we have now are significantly focused on showing you what the use of the novel glyphs does to your text color and texture.
• Should we make a paragraph/s for Pri and or SSA that addresses testing support for CCMP-delivered non-unicode letters? Since the time we made the paragraphs together, the importance of non-Unicode encoded letters that are heavily used and which are delivered via CCMP has been something I'm thinking more about. It might be that a list of these letters is also needed or maybe needed or instead of a paragraph.
Denis' reply:
Hi Eben,
Yes, Igbo and Zulu samples and updated Hausa, Fula and Xhosa samples would be useful. Akan and Dioula samples would be useful but they are relatively close to Bambara.
The GF Latin PriAfrican set targets Afrikaans, Hausa, Igbo, Luganda, Oromo, Swahili, Xhosa, Yoruba, Zulu, with the addition of Akan (Twi, Fante), Bambara, Dioula, Fula (Fulfulde).
Afrikaans, Oromo, Swahili, Xhosa and Zulu are already covered by GF Latin Core but Xhosa and Zulu stand out with lowercase-uppercase sequences in proper nouns as prefixes don’t get titlecased: for example "uSamuel" for "Samuel" or "AmaXhosa" for "Xhosa people". Note: Gaelic Irish also has lowercase-uppercase sequences. Octo-text already has a Xhosa sample, "nguVolume" and "zakwaYuda" seem to be helpful, maybe there could be more words that show potential kerning issues (words with aV, aW, aX, aY, eV, eW, eX, eY, oV, oW, oX, oY, especially with the symmetrical aYa, eWe, eYe, etc.).
There should definitely be Igbo and Yoruba samples as they use dots above and below letters. Igbo uses ị, and the sequence ịj would needs special attention. The other dotted letter sequences are less likely to be problematic as the letters are wider but it can still be useful. Igbo orthography normally doesn’t indicate tones with accents, so unfortunately examples of ị becoming dotless when combined with an accent are rare or limited to linguistic works like grammars or dictionaries.
The current Fula and Hausa samples do not include examples with ƴ. In Fula, the sequences ƴƴ and ƴi would need particular attention when the ƴ has a raised hook.
I think Akan, Bambara, Dioula are straightforward, but samples for Akan and Dioula would still be helpful.
Yes, samples should allow designers to test all the preferred variants or necessary substitutions and what the result should look like should be well defined.
I should clarify a couple of things about the ccmp feature as well as locl and calt features.
The ccmp feature is usually used to decompose or compose glyphs, for example composing eacute with
sub e acutecomb by eacute(although harfbuzz, coretext or directwrite do that automatically) or decomposing edotbelow withsub edotbelow by e dotbelowcomb.The locl feature is usually used to substitute glyphs by forms preferred in specific locales, meaning language systems, for example
sub Bhook by Btopbar.The calt feature can be used to substitute glyphs in some contexts, for example
sub i' @TopAccents by idotless, however because the calt feature is not applied consistently by several shapers or applications these substitutions often end up in ccmp instead.In any case, whether you mean the composition/decomposition glyphs necessary for functionality or preferred glyphs useful for better typography, I do agree the sample text should definitely have examples of those and descriptions of how they should look (or reference images):