Warning W1113 does not divulge which glyph has degenerated contours

Question

Warning W1113 does not divulge which glyph has degenerated contours

madig opened this issue 6 years ago · comments

Nikolaus Waxweiler commented 6 years ago

One of our fonts triggers this: W1113 Degenerated contour(s), Number of glyphs with the warning = 1. I guess the code to find these contours is at https://github.com/HinTak/Font-Validator/blob/master/Glyph/GErr.cs#L980, however:

I don't understand what the code does
It doesn't seem to use indGlyphOwner to note down the specific glyph index

I'd prepare a patch to fix 2., but the specifics string doesn't even seem to be used in the report?!

HinTak · Answer 1 · Mon Jun 11 2018 20:46:12 GMT+0800 (China Standard Time)

Sorry, the documentation is here: https://github.com/HinTak/Font-Validator/blob/master/NewHelp/W1113.htm Eventually it should be merged into here: https://github.com/HinTak/Font-Validator/blob/master/GenerateFValData/OurData.xml There are a few issues related to this, microsoft#60 microsoft#17 This issue probably should be re-opened: microsoft#15 I am surprised that it does not mention a glyph index. Can you send me a test file (if necessary, privately) to have a look?

Nikolaus Waxweiler · Answer 2 · Tue Jun 12 2018 00:07:45 GMT+0800 (China Standard Time)

(Sent to your at-users.sourceforge.net address)

HinTak · Answer 3 · Sun Nov 18 2018 10:04:51 GMT+0800 (China Standard Time)

Argh, I saw this problem while testing visual-studio-2008-built FontVal 2.1.3 with Microsoft's times - I 'll look into it at some point.

HinTak · Answer 4 · Tue Nov 27 2018 07:57:07 GMT+0800 (China Standard Time)

@madig : sorry it has taken a while. I think I can answer both question 1 and 2: the glyph id is kept by the caller, and also summary-filtered out by the caller, in OTFontFileVal/val_glyf.cs. What you want to do is to disable the filtering, which can be done by commenting out (I.e. prepend with "//") the line "GERR_CONT_DEGEN" in the struct m_namesInfoCnt near line 147.

As I said I saw it with Microsoft Times - there are 11 such glyphs, all in the u200x unicode range, for various sort of spaces. In all the cases, the glyph has a single contour with a single point, so the diagnosis is correct, though uninteresting. Removing the single contour would get the font smaller.

I had a quick look at some of the older reports I collected; many fonts have a non-zero summary count. I assume the small ones (10-ish or below) are the u200x spaces, but some fonts have summary counts in the 100's so cannot really be that.

You can have a go at commenting out that line and see what glyph id it says - I'd suggest start with a font with a high count.

I am okay pushing this change (the commenting) as 2.1.4, if you have trouble building/running with mono.

HinTak · Answer 5 · Mon Dec 03 2018 07:08:15 GMT+0800 (China Standard Time)

@madig : I tagged 2.1.4 a few days ago and just made the windows build and you can use that now. Wait a few days if you want/need the mac/linux builds.

HinTak · Answer 6 · Sun Dec 09 2018 08:18:58 GMT+0800 (China Standard Time)

@madig : the 2.1.4 mac/linux binaries with this fix are up . Would be interested in what you find - whether they are single point contours in the u200x unicode space range or something else.