RE: hmtx transform (was: Telcon reminder and agenda for Sep. 9)

Hi Jonathan,

On Wednesday, September 09, 2015 4:03 PM Jonathan Kew wrote:

> Re the 'hmtx' transform -- I think my inclination would be to just go with the simple "all or nothing" approach. IMO, it seems likely that in most cases, the tools and workflow used to create any given font will either produce consistent lsb and xmin values across all the glyphs, or won't ensure such consistency at all.

When it comes to lsb vs. xmin - the mismatch between them (if exists) stems from the differences between ink bounding box vs. bounding box for point coordinates. The decisions on where to place outline coordinate points are made by a human designer - it may be workflow specific but it's unlikely to depend on a particular set of tools. If a designer follows recommended practices and places the control points at the extrema  of the outline - the xmin and lsb are more likely to match but it's not a guarantee. And because it's typical for the multi-lingual fonts to contain a mixed bag of glyph outlines from different sources (such as e.g. CJK+ Latin fonts) - there is a definite possibility that while one set of glyphs may be well behaved other glyphs may not follow the same design approach.


> But if there's evidence that this isn't the case, and people want to pursue a more complex approach that allows us to distinguish the two arrays, I'm not opposed to considering that -- let's have a concrete proposal.

I'd very much appreciate your opinion on the approach I proposed yesterday: https://lists.w3.org/Archives/Public/public-webfonts-wg/2015Sep/0011.html

Thank you,
Vlad

Received on Thursday, 10 September 2015 17:58:07 UTC