A null offset is a valid value

[behdad]

In the fonttools subsetter I made it output null offset if a Coverage table is empty in MarkGlyphSets in GDEF. But ots fails on that.

Null offsets are perfectly fine according to the spec, and all implementations accept them. Yet another OTS stupidity :(.

ERROR: GDEF: Bad coverage location 0 for mark set 0
ERROR: GDEF: Invalid mark glyph sets
ERROR: GDEF: Failed to parse table
Failed to sanitise file!

[khaledhosny]

Do you have a test font (or how can I make one)?

[behdad]

@nyshadhr9 can you please provide this?

[simoncozens]

This (or something similar) also happens on attempting to sanitise Noto Sans VF's mark-to-ligature lookups.
https://notofonts.github.io/latin-greek-cyrillic/fonts/NotoSans/full/slim-variable-ttf/NotoSans[wght].ttf is an example.

[jfkthame]

In the fonttools subsetter I made it output null offset if a Coverage table is empty in MarkGlyphSets in GDEF. But ots fails on that.

Null offsets are perfectly fine according to the spec,

I'm not sure this is true.... where does the spec say this? It's not mentioned at https://learn.microsoft.com/en-gb/typography/opentype/spec/gdef#mark-glyph-sets-table, AFAICS.

(Note that MicrosoftDocs/typography-issues#828 (comment) and following comments seem to state the opposite, unless the specific field is explicitly documented as allowing null.)

[behdad]

Fair.

Let's look at it from this angle: All major implementations allow this.

[jfkthame]

That may well be so -- I haven't attempted to check. But Peter's comment at MicrosoftDocs/typography-issues#828 (comment) said:

Today, we have to assume that some implementations expect non-NULL offsets to these coverage tables (and, in fact, such do exist)

though I don't know what specific implementations he was referring to.

cc @PeterCon