- From: John Hudson via GitHub <sysbot+gh@w3.org>
- Date: Mon, 21 Nov 2016 20:11:56 +0000
- To: public-i18n-archive@w3.org
A few thoughts on this: 1. Ideally, underline position and thickness should be defined at the font level, with CSS only providing fallback default behaviour. There are, however, some issues with this: a) In OpenType fonts, values for underlinePosition and underlineThickness are stored in the '[post](https://www.microsoft.com/typography/otspec/post.htm)' table. There is one pair of values for the whole font, regardless of how many different scripts are supported, and currently no way to specify different values for different scripts or language (writing) systems. b) font tools tend to provide a default setting for these two values, and a lot of font makers appear not to alter the default setting, with the result that the data may not produce best results. I think both of these issues could be addressed by defining localisable underline thickness and position values, probably as an extension of the existing but little-used 'BASE' table. This would not only provide layout software with appropriate values for different scripts and writing systems, but the newly defined data structure may well encourage font developers (and font tool makers) to pay more attention to this aspect of font production. This could also assist CSS in determining whether the font data is reliable. [These comments also apply to the strikethrough height and thickness values, currently implemented in the 'OS/2' table, and similarly under-localised.] 2. A lot of writing systems have conjunct forms, below marks, and/or vertical stacks that cut through the height of an underline. In addition to determining the appropriate height and thickness of underlines, it is also desirable to determine the preferred behaviour of underlining with descending glyphs, which might also differ depending on writing system. This is complicated by the fact that the depth of a given sequence of characters in a complex script is font-dependent, and cannot be determined or even reliably guessed at from the character string. -- GitHub Notification of comment by tiroj Please view or discuss this issue at https://github.com/w3c/ilreq/issues/28#issuecomment-262053066 using your GitHub account
Received on Monday, 21 November 2016 20:12:04 UTC