Isolated lam-alef ligature is not formed in Noto Kufi Arabic

[khaledhosny]

Describe the bug
The isolated lam-alef ligature is not formed in Noto Kufi Arabic

To Reproduce
Check:
https://fonts.google.com/noto/specimen/Noto+Kufi+Arabic?query=noto+kufi&preview.text=%D9%84%D8%A7%20%D8%A8%D9%84%D8%A7&preview.text_type=custom

Expected behavior
Lam-alef ligature is a required ligature and should be formed all the time.

Screenshots
Rendering on fonts.google.com:

Expected rendering:

[simoncozens]

This appears to be a subsetter issue, not a font or website bug.

The variable font contained in this repository forms the ligature OK, but when subset for delivery by fonts.google (i.e. even if you hit the "Download family" link and look at the variable font you get there), the ligature goes away.

In the original font, the ligature rule is in two lookups, LigatureSubstitution9 which is attached to the rlig feature, and LigatureSubstitution10 which is attached to the dlig feature:

lookup LigatureSubstitution9 {
    lookupflag RightToLeft IgnoreMarks;
    ;
    # Original source: 8
    ...
    sub uniFEE0 uniFB51 by lam_alefWaslaar.fina;
    sub uniFEE0 uni0672.fina by lam_alefWavyhamzaabovear.fina;
    sub uniFEE0 uni0673.fina by lam_alefWavyhamzabelowar.fina;
    sub uniFEDF uniFE8E by uniFEFB.isol;                                            # HERE (rlig)
    sub uniFEDF uniFE84 by uniFEF7.isol;
    ...
} LigatureSubstitution9;

lookup LigatureSubstitution10 {
    lookupflag RightToLeft IgnoreMarks;
    ;
    # Original source: 9
    ...
    sub uniFEE0 uni0675.fina by lam_highhamzaAlefar.fina;
    sub uniFEDF uniFE8E by uniFEFB.isol;                                            # ALSO HERE (dlig)
    sub uniFEDF uniFE84 by uniFEF7.isol;
    ...
} LigatureSubstitution10;

In the subsetted font, LigatureSubstitution9 no longer contains the ligature rule (or lam-alef-hamza etc):

lookup LigatureSubstitution9 {
    lookupflag RightToLeft IgnoreMarks;
    ;
    # Original source: 8
    sub uniFEE0 uniFB51 by lam_alefWaslaar.fina;
    sub uniFEE0 uni0672.fina by lam_alefWavyhamzaabovear.fina;
    sub uniFEE0 uni0673.fina by lam_alefWavyhamzabelowar.fina;
    sub uni06B5.medi uniFE8E by lamVabove_alefar.fina;
    sub uni06B6.medi uniFE8E by lamDotabove_alefar.fina;
    ...
} LigatureSubstitution9;

Although it is still present in the dlig-attached lookup:

lookup LigatureSubstitution10 {
    lookupflag RightToLeft IgnoreMarks;
    ;
    # Original source: 9
    sub uni0627 uniFEDF uniFEE0 uniFEEA by uniFDF2;
    sub uniFEE0 uni0672.fina by lam_alefWavyhamzaabovear.fina;
    sub uniFEE0 uni0673.fina by lam_alefWavyhamzabelowar.fina;
    sub uniFEE0 uni0675.fina by lam_highhamzaAlefar.fina;
    sub uniFEDF uniFE8E by uniFEFB.isol;
    sub uniFEDF uniFE84 by uniFEF7.isol;
    sub uniFEDF uniFE88 by uniFEF9.isol;
    sub uniFEDF uniFE82 by uniFEF5.isol;
    sub uniFEDF uniFB51 by lam_alefWaslaar.isol;
	...
} LigatureSubstitution10;

[khaledhosny]

I don’t know what subsetter is used, but I can’t reproduce this with either hb-subset or fonttools subset. cc @garretrieger

[garretrieger]

Don't think this is subsetter related. I believe the issue is that the copy of Noto Kufi Arabic we are serving on fonts.google.com is outdated. I can reproduce the issue on our unsubstted internal copy of the font. I'll get our copy updated.