Thin spaces

[quasicomputational]

Thin spaces (U+2009 THIN SPACE and U+202F NARROW NO-BREAK SPACE) are pretty nice as a way to separate digit groups, but they're not currently present in Source Serif Pro. Could they be added, please?

[Moilleadoir]

Since support for it is so bad, I doubt anyone actually uses it, but U+202F NARROW NO-BREAK SPACE has a potential use for punctuation spacing in French. See HTML authoring in French.

[throwaway571]

Since support for it is so bad, I doubt anyone actually uses it, but U+202F NARROW NO-BREAK SPACE has a potential use for punctuation spacing in French. See HTML authoring in French.

I can confirm that I use this character whenever possible, both for French punctuation and for spacing before unit symbols:

La vitesse à 30 km/h sur plus de 80 % du réseau nantais a un double objectif d’après la maire de Nantes Johanna Rolland : « Renforcer la sécurité des piétons comme des cyclistes, mais aussi de réduire les nuisances sonores pour une meilleure qualité de vie. »

[frankrolf]

Hoping that’s a reasonable assumption:
I am thinking of making the THIN SPACE and NARROW NO-BREAK SPACE the same width – corresponding to the calculations here: https://helpx.adobe.com/indesign/user-guide.html/indesign/using/glyphs-special-characters.ug.html (⅛ em).

[quasicomputational]

I had the impression that they were usually about 1/6 em, but weren't necessarily constant (e.g., could be stretched for justification). I have found 1/9 em to be too narrow for digit group separation (mhchem/MathJax-mhchem#18 has some pictures), and I think 1/8 em would possibly also be too narrow. Is it possible that that document's using a slightly different definition of 'thin space'?

[frankrolf]

I would like to keep the impact on users of InDesign low, that’s why I will likely follow the factors on that help page.
I have reconstructed the exact distances in a test file:

It turns out that InDesign actually uses the width of the normal space character for U+202F NARROW NO-BREAK SPACE, which is probably not the original intent of “NARROW”:

I decided to deviate from InDesign’s calculation for that one space, and make it equal to U+2009 THIN SPACE.

What’s “too narrow” is of course a matter of personal preference, but I think the thin space works as a number separator:

[frankrolf]

for posterity – here are the factors used:

def make_space(name, unicodes, factor):
    g = f.newGlyph(name, clear=True)
    g.width = int(round(1000 * factor))
    g.unicodes = unicodes


spaces = [
    # name, unicodes, factor
    ('enspace', [0x2002], 1 / 2),
    ('emspace', [0x2003], 1),
    ('threeperemspace', [0x2004], 1 / 3),
    ('fourperemspace', [0x2005], 1 / 4),
    ('sixperemspace', [0x2006], 1 / 6),
    # ('figurespace', [0x2007], None)  # = zero
    # ('punctuationspace', [0x2008], None)  # = semicolon
    # ('narrownbspace': [0x202F], None),  # Indesign assumes = space, which is odd
    ('thinspace', [0x2009, 0x202F], 1 / 8),
    ('hairspace', [0x200A], 1 / 24),
    ('zerowidthspace', [0x200B], 0),

]

# write lines for GOADB
for friendly_name, unicodes, _ in spaces:
    if len(unicodes) > 1:
        final_name = friendly_name
        override = ','.join([f'uni{uc:04X}' for uc in unicodes])
        line = '\t'.join([final_name, friendly_name, override])
    else:
        final_name = f'uni{unicodes[0]:04X}'
        line = '\t'.join([final_name, friendly_name])
    print(line)


a = AllFonts()
for f in a:
    for gname, unicodes, factor in spaces:
        make_space(gname, unicodes, factor)
    f.save()
    f.close()

[frankrolf]

Various spaces are part of Source Serif 4.004: https://github.com/adobe-fonts/source-serif/releases/tag/4.004R