Any reason why winAscent + winDescent can’t be smaller than the UPM?

[schriftgestalt]

afdko/c/makeotf/makeotf_lib/source/hotconv/hot.c

Lines 716 to 721 in 5778828

	intLeading = font->win.ascent + font->win.descent - font->unitsPerEm;
	if (intLeading < 0) {
	/* Avoid negative internal leading */
	font->win.ascent -= intLeading;
	intLeading = 0;
	}

This might be a good check for auto calculated winAscent + winDescent values. But quite confusing for manually set values.

[josh-hadley]

This code has been around for quite a while, probably with the same logic since the beginning. I don't think anyone is still around who was involved in the decisions on why it's done. I suspect it's rooted in the OT spec recommendations around metrics and BTBD or something along those lines.

I would be open to the idea of changing this to detect-and-warn, without modifying any values. That way, anyone who is reliant on this behavior (knowingly or not) would at least be notified and make modifications to modify the values, if desired.