[OpenFontLibrary] Status report about the new OFLB

[Jon Phillips]

This is so great Dave and co! The site looks amazing and I'm uber
excited about the analysis tool. I'll keep on OSUOSL admins to get our
VM setup so we can host all this great work!

Cheers!

Jon

On Thu, 2009-01-01 at 21:04 +0000, Dave Crossland wrote: 
> Hi All,
> 
> Progress on migrating OFLB to the new site has been delayed by (a)
> moving to a new virtual server host thank to OSUOSL.org and (b) not
> yet having the font list - http://openfontlibrary.fontly.org/files -
> working well. I hope both these issues will be resolved this month :-)
> 
> We do have really nice INTERACTIVE font previewing thanks to Ed Trager
> though - click the book icons at
> http://openfontlibrary.fontly.org/files/admin/6 for example - and Ed
> is now working on a key feature for community development of fonts -
> showing how much language coverage a font has. For the easter (well,
> summer? ;-) I hope that we'll have this up and running :-)
> 
> Here is an email from Ed today about the new feature that I thought we
> should really discuss on list :-)
> 
> Happy new year!
> 
> Dave
> 
> ---------- Forwarded message ----------
> From: Ed Trager <ed.trager at gmail.com>
> Date: 2009/1/1
> Subject: Font Analysis Program :: Thoughts on Fonts That Provide Only
> Partial Orthographic Coverage
> 
> Hi, Dave and Ben,
> 
> Happy New Years!
> 
> I'm making progress on the font analysis program and it is quite
> interesting to see the kinds of coverage that turn up in real-world
> fonts.
> 
> One interesting (but not unexpected) phenomenon is that fonts often
> contain significant but still *incomplete* coverage for certain
> orthographies.
> 
> For example, here is the current "DEBUG" mode output for Aboriginal
> Sans, a font for Latin-based native American language orthographies by
> Chris Harvey of LanguageGeek.com:
> 
>   Basic Latin is supported!
>   Western European is supported!
>   Euro is supported!
>   Catalan is supported!
>   Baltic is supported!
>   Turkish is supported!
>   Central European is supported!
>   Romanian is supported!
>   Vietnamese is supported!
>   Dutch is supported!
>   Afrikaans is supported!
>   Pinyin is supported!
>   IPA FAILED with 84 hits on 86 tries.   <== NOTICE THIS
>   Latin Ligatures is supported!
>   Common Name :Aboriginal Sans
>   Native Name :
>   Sub Family  :Regular
>   Has Vertical:0
>   Style       :normal
>   Weight      :normal
>   Fixed Width?:0
>   Fixed Sizes?:0
>   Num Glyphs  :5084
>   Num Chars   :4975
> 
> So we see that "Aboriginal Sans" actually does provide almost complete
> coverage of the International Phonetic Alphabet (IPA), excepting 2
> characters. That hardly represents a "failure" to cover IPA. I have
> not yet investigated which two characters are missing -- but we can
> infer that they must be rarely-used ones, or at least rarely-used in
> Indigenous American language orthographies.
> 
> Significant but still incomplete coverage is notable also in CJK
> fonts.  Here's an example from an older copy of the Open Source
> Firefly Sung Chinese font:
> 
> Basic Latin is supported!
> Euro is supported!
> Pinyin is supported!
> Simplified Chinese FAILED with 3499 hits on 3500 tries.
> Traditional Chinese is supported!
> 
> Again, "failure" due to one missing character out of 3500 does not
> really mean lack of coverage -- most likely the missing character
> represents one of those "glyph variants" that should never have been
> encoded as a separate character in Unicode, but got there anyway.
> 
> Properly defining "orthographic coverage" in a Chinese font is in
> itself an interesting problem!  The various national and international
> standards such as BIG-5 and GB enumerate thousands more Chinese
> characters than are actually used by literate educated people on a
> daily basis.  Reading a modern Chinese newspaper requires knowledge of
> between 3 to 4 thousand characters.  I forget how many BIG-5
> enumerates, but I think it is on the order of 20,000 characters -- in
> other words, a lot more than most people would need.  Unicode
> enumerates even more, especially when the (plane 1) HKSCS is included.
> 
> So what I decided to do for Chinese was use a list of the most
> frequent top 3500 simplified and top 3500 traditional (there is much
> overlap between these sets).  I think this is a reasonable approach.
> For example, Chinese and Japanese "art" fonts used for advertising and
> graphic design are known to only contain the more common characters
> and the abstruse, archaic, and rarely-used technical characters are
> left out.  We can anticipate similar phenomena during the development
> process of future Open Source CJK fonts.
> 
> On the web site, we would like to be able to display summary
> orthographic coverage results compactly and succinctly.  The idea of
> using a set of little graphical icons has been discussed, and I
> created demonstration artwork for 15 Latin orthographic categories.
> 
> However, currently those icons are only useful in representing boolean
> states: "covered" or "not covered".
> 
> Perhaps we can consider adding color to the icons.  I would propose
> having not more than 3 states:
> 
>   1. Full Coverage
>   2. Partial Coverage -- more than 50%
>   3. Partial Coverage -- less than 50%
> 
> (There is actually a 4th state, "No icon displayed at all" meaning "no
> coverage at all" for a given orthography).
> 
> In terms of colors, we might initially think "Green--Yellow--Red" as
> used in traffic lights.  But as I am often reminded at the Kellogg Eye
> Center where I work, a significant percentage of the population,
> especially men, are red-green color blind. The rate is something like
> 8% of the male population.
> 
> So green and red are poor choices.  Blues, yellows, and yellow-oranges
> are much better choices -- and this is undoubtedly one reason why we
> see the predominance of blue in web site color schemes.
> 
> So what if we used the following:
> 
>   1. Icon with BLUE                       background  -- FULL Coverage
>   2. Icon with YELLOW-ORANGE background  -- > 50% Coverage
>   3. Icon with GRAY                      background  -- < 50% Coverage
>   4. NO ICON                                                     --
> NO coverage for this orthography
> 
> We could even have tooltips so if you hover over the icon with the
> mouse, it could indicate the actual % coverage.
> 
> I've attached a quickly-done mockup of some icons.
> 
> Let me know what you guys think about this kind of issue.
> 
> Best Wishes - Ed
-- 
Jon Phillips 
http://rejon.org/
San Francisco + Beijing
GLOBAL +1.415.830.3884 - CHINA +86.1.360.282.8624
IM/skype: kidproto - Jabber: rejon at gristle.org
BIO http://rejon.org/bio - CV http://rejon.org/bio/cv