Closed chrissimpkins closed 6 years ago
@jublo missing from general punctuation glyphs that are necessary to complete the glyphs that are not spaces (U+2000...U+200A), not format characters (U+200B...U+200F), not "other" (U+2060...U+206F)
U+2022
- bullet (will need to check subsetting list)U+2044
- Fraction Slash@jublo
Missing in currency set:
U+20AF
- Drachma (missing italic set only)U+20B6
- Livre Tournois Sign (13th-18th century French symbol)U+20B7
- Spesmilo SignU+20BA
- Turkish Lira SignU+20BB
- Nordic Mark SignU+20BC
- Manat SignU+20BD
- Ruble SignU+20BE
- Lari SignU+20BF
- Bitcoin Sign@jublo see David's manual analysis in https://github.com/source-foundry/Hack/issues/286#issuecomment-327592158 above. He did not find some glyphs on manual inspection of the regular web font set that are showing up in your visual proofing images. Are we certain that we are displaying glyphs that are present in the web fonts (complete and subset) and not fallbacks in some cases?
@jublo @chrissimpkins I manually inspected via Chrome Dev Tools, and checked whether a certain glyph was rendered via the webfont, or from a fallback. What struck me, was that some of the missing glyphs were displayed by different fallback fonts that I had not selected as such. Jublo, how does your tool determine whether it's been displayed as Hack or not?
What struck me, was that some of the missing glyphs were displayed by different fallback fonts that I had not selected as such. Jublo, how does your tool determine whether it's been displayed as Hack or not?
@burodepeper @chrissimpkins I can confirm this behaviour right now – U+2055 is rendered as Monaco transparently, but I have not indicated that. This is serious and I need to inspect it.
@jublo thanks for looking into it JM. Perhaps fallback and metrics are the same between the two so detected as present in test sets?
May be in a browser default fallback definition/setting rather than in CSS?
@chrissimpkins I need to find out – but not today.
@jublo I will take a look through the web font subsets to confirm all actual glyphs that are included so that you have a baseline to work from with the bugfix (or new tool if you go that route)
@chrissimpkins U+2055 is affected in any case.
@burodepeper @jublo
Here are the missing glyphs as defined in the Hack source (i.e. before subsetting):
ASCII: None , set is complete
Latin-1 : None , set is complete
Latin Ext A:
General Punctuation:
Currency:
Next step will be to evaluate the web font subsets to confirm that they contain all of the expected glyphs in these sets excluding those that are missing from the source (above list).
We are in good shape with the subset web fonts. I reviewed every set manually and here is where we are:
ASCII: all sets complete
Latin-1: bold italic required minor fix https://github.com/source-foundry/Hack/commit/3adbae7184ff61fb13f75fc5d2042b8d653ae126, now all sets complete
Latin Ext A: consistent with glyphs that are missing from the set in the source
General Punctuation: in addition to the glyphs that are not currently included in the Hack sets, the following were missing from the subsets:
None of these are critical glyphs and we can troubleshoot as part of the v3.x line of work. Likely will be an error in the naming within the subset definition files. Not something that IMO should hold up release at this stage.
Currency: consistent with glyphs that are missing from the set in the source
The additional glyph numbers (beyond expected for the defined sets) are due to inclusion of additional fractions, additional subscript, additional superscript glyphs that are likely pulled in from the features defined in the fonts. This is not problematic, rather this is desired behavior of the compilation tool.
Let me know if you have any concerns. My feeling is that we are good to go from the web font build aspect. Still need to address the CSS files (require revisions for the new release files that we are using) and the package.json file (to support CDN releases of the web fonts via npm >> jsDelivr). This should complete all outstanding v3.0 work for the web fonts (there will be some web font documentation updates which I am not considering as part of this work).
@chrissimpkins Do you judge the glyphs missing from the main fonts to be a 3.0 blocker?
@jublo Don't believe that anything currently missing is critical. These are things that we should fix over time, but are not in high priority category for fixes. My suggestion would be to aim to add all missing glyphs from https://github.com/source-foundry/Hack/issues/286#issuecomment-332055383 and address the bug in the general punctuation set (already designed and present in source, not in subsets) from https://github.com/source-foundry/Hack/issues/286#issuecomment-332395527 as part of our v3.x work. It would be nice to have complete sets across all of these targeted sets (if this is what we are going to use for web font subset targets...).
Would be worth continuing the discussion about what we feel is important to support in the subsets. The sets (derived from what we include above + other sets) defined for Google Fonts in https://github.com/davelab6/pyfontaine/tree/master/fontaine/charsets/internals/google_glyphsets may be worth more conversation. Let's discuss and then begin to add these glyphs as part of upcoming work on the fonts.
@jublo is adding missing glyphs from the Latin Ext A set. Coming in as PR now
JM PR x 3 merged to dev
branch. One outstanding glyph design to complete the Latin Ext A set is still under discussion.
@chrissimpkins Are you sure that these are to be added? They’re spaces and 0-width chars?
U+200b...U+200f (missing in all sets)
This also applies to U+2028…U+202E.
Would love to get insight on U+2038
.
Are you sure that these are to be added? They’re spaces and 0-width chars? This also applies to U+2028…U+202E.
Agree we do not need to add zero width glyphs. I wasn't aware of this when I reviewed the code ranges. Thanks @jublo
Would love to get insight on U+2038.
@jublo By insight do you mean is it necessary to include it? Will need to look through these missing glyphs to see what they are. I examined Unicode code points to determine missing values when we reviewed this. Are you trying to complete the general punctuation set for v3.001?
I would like to get an opinion for my pasted design. :)
lgtm
Reading up on this. It appears that the caret shape at this point (c/w the ASCII ^
) is meant to align with the baseline as a proofreading insertion character. Agree?
‸
vs. ^
U+2038 ‸
U+005E ^
Want to open a new issue report for this? We are discussing on a closed thread...
woff is supported in all modern browsers. TTF, EOT, and SVG web fonts are no longer recommended.
Suggest that we build woff (+/- woff2) files only for web font releases.
build-woff.sh
&build-woff2.sh
&build-subsets.sh
$ make webfonts
)