Changelog

Changelog since this site went live on 2024/08/11

Please note that I often did not record revisions to the HTML, nor minor
corrections to templates or other supporting files.

May 2025

Unless anyone reports missing files to me in the next few days, I do
not intend to make any further changes.
Since things were quiet here, I looked at a comment from Zdenek
Wagner about some fonts having incorrect spacing values. I decided to
take a look at using LuaLaTeX for Vollkorn whcih has an Old-English or
Sami shape for 'Eng' (my preference) and supports Norrthern Sami. To
my immense surprise, not only does Vollkorn work well with normal text
using polyglossia (minor tweaks for Vietnamese tone marking layout),
but the small caps also work well with polyglossia and I can show the
small cap Vietnamese tone markings in an "adequate" fashion.
For my own interest I created lipsum-ukrainian files using LuaLaTeX to
show the different variations of some letters when used in Cyrillic italics.
I discovered that LuaTeX will not accept a digit in a font identifier, so
SourceSans3 is rejected but SourceSans is accepted, and of course the
digit is required in the fontconfig name of 'Source Sans 3'. Getting
polyglossia to NOT change all the Ukrainian text to the default font was
interesting.
Updated WenQuanYiZenHei.
Updated NotoSans, NotoSerif and NotoSerifCJKkr.
I eventually gave up trying to show small cap Vietnamese vowels in the
tone markings, at best the layout was very ugly, at worst some lines used
lowercase.
Found the fonts-alee Korean fonts I had intended to update. In January or
February I found a link to a github site, but I have lost that link and cannot
find the site. Found them at Salsa.debian.org,but this was an extremely
painful revision - Guseul is actually unusable, and Bandal and Bangwool
make a mess of the contents PDFs in libreoffice.
Updated to AdwaitaSans and AdwaitaMono (I list Sans first, Mono attempts
to match it). Unfortunately, this has reduced my faith in LuaLaTeX (I am still
on a 2024 build and lack time to build a newer BLFS and then to build 2025
from source) : In Sans the font justifies, except in "Quotations" mentioning
Spanish punctuation and the Tironian shorthand for 'et' sometimes seen on
Irish web pages. In these cases, LuaLaTeX breaks lines in odd places, even
after forcing a newline. I had to replace all spaces and non-breaking spaces
with hspace directives in millimetres. For AdwaitaMono the font does not
justify at all and I gave up trying to show Article 1 in Northern Sami (neither
font has appropriate Eng glyphs for Sami, using the African style). For the
Mono font, similar use of hspace, but with much tinier measurements,
allowed me to sort out the Vietnamese tone markings and the 'Quotes'
items.
The new process is now "good enough". For fuller details see
current-process.html
Updated PDF-contents files (because of checking combining codepoints) for
DejaVuSans, FreeMono, FreeSans, FreeSerif, NotoSans, NotoSansKannada,
NotoSansKhmer,NotoSansSymbols, NotoSansTibetan, NotoSerif, Padauk,
Scheherazade.
Updated unicode-blocks, added mightcombine and newgenerate scripts to
files/tools/
Moved the old unicode-blocks and generate-all-characters files to
files/tools/old-unicode

March 2025

Meanwhile, I worked on preparing the languages file for Adwaita Sans and
discovered that my previous use of negative spacing ('\|') to sort-of align the
Vietnamese tone examples no-longer worked (this is still TL2024) and in the
end I updated the font-languages-new template to use '\hspace' between
the letters. For that font it works well, so arguably I am now a typesetter.
During that time, the new Gnome Adwaita fonts were released. I discovered
that I can document the language support of the Variable Adwaita Sans font
if I use LuaLaTeX. Unfortunately, listing the contents showed some issues in
my generate-all-characters script: These fonts use new blocks assigned in
Unicode-16.0 which I updated to, but after looking at a vertical 'smudge' in
one of the Cyrillic extensions I discovered that many Combining codepoints
are scattered through blocks not labelled as Combining. Attempting to work
around that has not yet succeeded, further analysis is stalled, hoping I can
complete that (not lookign good at the moment).
I stepped away from this to check the TL2025 pretest offering.

February 2025

Renamed Korean name files in files/korean and Hangul-Hanja.html to
fontname-forenames and fontname-surnames because the existing
naver-fontname and Korean-surnames-fontname were unwieldy and
unmanageable when uploading in filezilla.
Really fixed KR-punct files.
More updated or revised Korean fonts: NanumGothic, NanumMyeongjo,
UnBatang, UnDotum, UnGraphic, NotoSansCJKkr, NotoSerifCJKkr. For
NotoSerifCJKkr I updated the version; for the moment updating the other
NotoSerifCJK fonts is a low priority - I do not know what changed, and the
subset of codepoints returned by my defective script is unchanged.
Noticed that downloading Noto fonts (and therefore other fonts from
Google) now requires clicking on 'Get font' at bottom right of the window,
so altered the comment for Google fonts.
Updated Noto Sans to check my LuaLaTeX template for files with more
general coverage. Added an -sc template.
Altered the LuaLatex languages template to use page headings,
and revised the Baekmuk languages files for that.

January 2025

Starting to update Korean fonts:
Getting ready to look at Korean fonts, I discovered that the Baekmuk,
Nanum and Unfonts links no-longer worked. Updated.
After the previous update, I tried adjusting one example Chinese font
variation to get 3 glyphs on the last line, using spaces to pad out the
previous line. But I was unable to get the padded line to match the
length of the original line and there were no useful messages. I conclude
that adjusting Chinese text should be left to those who can read and
reword it.
I have now prepared some Chinese-variant files, both to compare how
different fonts render the same text, and to compare how the equivalent
text is rendered in CN, TW and HK variants. This has caused me to reject
XeLaTeX in favour of LuaLaTeX (XeLaTeX is unmaintained). When I get to
revising Korean fonts I might continue to use XeLaTeX, but I now
recommend anyone doing similar font exercises should use LuaLaTeX.
Follow the links from cjk-comparisons.html to get to these and all other
CJK-specific files.
While I was preparing files re Hangul coverage, I noted that fontconfig
happened to choose Noto Sans Mono CJK HK on my British system, probably
by random chance. Then I eventually made progress in preparing files to
compare the three common variaties of Chinese (Simplified, Traditional,
Cantonese for Hong Kong) and was shocked to discover I had not analyzed
that font. Now added.

December 2024

Corrected titles in some lipsum-TW files which I seem to have already added
to my local git tree. Added lipsum-HK files (only sans and serif, few fonts, I
have include the kai with the other serif fonts. Added lipsum-??-new.tex
templates for CN, HK, JP, TW in case I ever add more fonts for those
languages.
Created the example Han glyph files for CJK-HK, CJK-JP, CJK-TW.
Revised the Cantonese (HK) fonts, added Chiron Hei HK and Chiron Sung HK
together with their Text variants. (Han codepoints in slightly reduced width).
Uploaded an HK-punct PDF showing Chiron Hei HK which has the fullest
overage.
Revised UMingTW languages file, I had labelled it as coming from ukai.ttc.

November 2024

I have revised the lipsum-TW-sans fonts to include the GenYo Gothic
fonts. This means that I now have Traditional Chinese Sans fonts with
mid-level ellipses.
Various text corrections in languages files for Traditional Chinese fonts.
I had not discovered the GenYo Gothic (i.e. Sans) fonts which are in a
separate repo. Added these for the second batch of November updates.
Created Traditional Chinse lipsum-TW- files for kai, sans and sung.
I had moved the CJK-CN-*.tex files to files/cjk-tex/ but left older copies in
files/tools/templates. I have now removed those older copies. I intend to
make similar comparisons for other CJK languages, but at the moment
that is a low priority.
For TW fonts I also show the local Minnan dialect (as usual, from Omniglot).
For reasons I do not begin to understand XeLaTeX formats the lines of that
Article 1 with one less ideogram (33 instead of 34) compared to CN, HK, JP,
and TW Mandarin.
Updated Traditional Chinese (TW) fonts, adding some GenYo and HanWang
fonts. I looked at, but ignored, the GenYo Japanese fonts because I could
not see the point, nor their differences from Source Han Serif (Noto CJK).
I also looked at the Glow Sans fonts - a programmed modification of
Source Han Sans to provide wider or narrower forms. For comparing how
different fonts appear in my lipsum files I need to use the Normal width,
and for that the only difference in the CJK codepoints is that Source Han
has wider coverage (some glyphs could not be adapted, the project is no
longer maintained). The ellipses in Glow Sans are at mid level, so I tried it
because none of the other libre Traditional fonts have mid-level ellipses.
But it omits many codepoints used throughtout my tests for language
coverage, even in the zhlipsum 'trad' text which all other Traditional fonts
manage, and also in my own prepared texts for lipsum files. For me it is
not usable.
I have now created lipsum files for Japanese Sans and Serif fonts. For
consistency with my working files and with Chinese variants I have
named these as -JP-. Unfortunately, one of the Serif fonts showed that
my attempt to use seven full lines to help show missing glyphs was not
effective (they were relaced by normal width spaces, XeLaTeX then
adjusted the line to full width).
I moved the extra Japanese fonts to a permanent location on my system.
For Shippori Mincho I had forgotten to mention the odd full-width
exclamation-mark, so the PDF has been updated for that. For the Zen
fonts only the fontfile locations in the TeX languages files have changed.
After looking at the newer Japanese fonts I went back to Ume Hy Gothic
and changed that to the O5 (medium weight) version - at small sizes some
Kanji look much darker than others, but at large sizes the weights are less
uneven.
Added more Japanese fonts: IPAMincho (I had ignored this until now in
favour of IPAexMincho which is newer and not monospaced Latin, but I
have now noted that IPAMincho can be a fallback, in certain circumstances,
for Japanese Serif), Shippori Mincho, Zen Kaku Gothic New, and Zen Old
Mincho. The last three have multiple weights, I have used the weights
which look most readable to me for black text on white.
Revised the Droid Sans Fallback and IPAex fonts again, to fix errors or
unfortunate paging.

October 2024

Updated the IPAex fonts to v004.01, revised the Kochi fonts, updated the
sawarabi fonts which are now at Google and named SawarabiGothic,
SawarabiMincho and relicensed.
Revised, again, those Japanese fonts I had revised in September. Also
odosungmono which I had overlooked. Updated XX-test and XX-punct files
(.tex and .pdf) in tools/templates/ and cjk-tex/. Unfortunately, the HK test
file often needs to be looked at line-by-line to see if all the expected glyphs
are present.
Revised the Simplified Chinese (CN) fonts, in particular correct coverage of
which CJK languages are supported, comment on any issues with the CN
punctuation, and correct the Article 1 text to match what was current at
Omniglot (slightly longer than what I previously had). At the moment I
cannot update the old fireflysung datails.
Revised the lipsum-CN-* files. More text, comment on low ellipses.
Updated the languages templates to use \raggedright for languages such as
Azeri or Adyghe where I lack hyphenation patterns. I only expect to use the
revised templates when updating CJK fonts, few of those will be affected.
Created lipsum-CN-{kai,sans,sung} files to replace the old
lipsum-simplified-chinese files. Unlike Japanese, Chinese has two styles of
font which I regard as serif.
I realised that the identifiers for the Noto Serif CJK fonts in CJK-CN-kai and
CJK-CN-sung were the numbers for the Noto Sans CJK fonts. Corrected.

September 2024

Updated to fonts-horai-umefont_670. I now show the Ume Hy Gothic font for
item 1.131 because it includes blank non-breaking spaces and an identifier for
missing codepoints. I also attempt to explain the available variations, and the
problems with the Ume fonts for non-Japanese languages.
Reworded tools.html and cjk-comparisons.html.
Updated VL-Gothic to the current version.
Looking at older CN fonts, I now find that JP and TW are not adequately covered.
I have revised the languages files for the following: fireflysung, odokai, odosung,
odosungmono, UKaiCN, UMingCN.
I started to look at other Japanese fonts, then noticed some fonts admit to not
having as many Kanzi glyphs as desired. I eventually created templates for
testing the coverage of CN, JP and TW for a font. I also discovered that,
particularly for Japanese, variant shapes of some Han glyphs might be
preferred. Looking at google translate while creating these files, I finally
understood why CJK languages usually omit 'gibberish' dummy text: almost any
combination of syllables can have a somewhat-reasonable translation. These
templates are in files/tools/templates and the comments mention where I got
them from.
Karl Berry suggested my usage of 'sc' for both Small Caps and Simplified
Chinese could cause confusion. He suggested zh-CN. As far as I can see, only
the files comparing shapes of codepoints are affected. I have renamed these to
CJK-CN-kai, CJK-CN-sans, CJK-CN-sung.
Updated the Hanazono fonts - although Debian has a newer version than what I
used to use, the latest version, v8.030, is at github v8.030. That changed the
fontconfig names to Hanazono Mincho A,B and added a C version (item 3.022 in
my table) to cover later CJK extensions. Comments on how to use the B and C
fonts are in 'HanaMinA-languages.pdf'.
Created comparative files showing certain codepoints for Simplified Chinese:
CJK-sans-sc, CJK-serif-sc-sung (includes Fang), CJK-serif-sc-kai.
These all start by showing glyphs from Noto and WenQuanYi Zen Hei and are
linked from cjk-comparisons.html. Some of those codepoints are not used in
Simplified Chinese, I intend to do other CJK languages using similar files.
Revised gbsn00lp, gkai00mp, UKaiCN, UMingCN, template languages-full.tex.
Revised odokai to include angle brackets, fixed typo in odosung{,mono}.
Revised the lipsum-greek-sans and lipsum-greek-serif files.

August 2024

Fixed the Junicode languages files, Greek had accidentally been omitted.
I have now revised the 'PDF-substitutes' files.
Attempted to simplify CJK descriptions: replace Pan-CJK by the actual language,
consistently name as (Monospace) Language Sans/Serif. For Chinese recognise
that Serif is split into Sung, Kai, Fang.
Revised odosung and odosungmono to include angle brackets.
Fixed an error in the "identify CJK language part of languages-full.tex:
I had shown U+5377, should have been U+537F as in Japanese part.
For small caps in Azeri and Turkish I had wrongly added the combining dot above
to a normal lowercase i instead of to a dotless i. For most fonts this makes very
little difference to my conclusions about whether or not they are usable for Turkic
languages. But I noticed oddities in hyphenation for Linden Hill and OFL Sorts Mill
Goudy which suggest that at least those two fonts may give problems with
hyphens after dotted i in Turkish small caps using XeLaTeX and polyglossia. For
the fonts which support small caps in Azeri (for which I have no hyphenation
patterns) the same may be true if they are used for Turkish.
Revised the languages files for all Noto CJK fonts, both WenQuanYi Zen Hei, and
Droid Sans Fallback to better show their coverage of the Korean Hanja I know about
and to include angled brackets and double angled brackets for quotations.
Added Hangul-Hanja files for Droid Sans Fallback. Some of the many missing
codepoints in that highlighted duplications in my earlier uploads and the template,
for 'ga' and 'joon' . I have fixed the template but will not revise the other files, the
additions are benign and I'm sure there are other duplications.
Added CJK-sans-sc (Simplified Chinese, Sans) files showing Han glyphs which vary
across languages. Unfortunately, the differences seen in Serif fonts are not always
visible in Sans fonts.
Added a page for linking to comparisons of CJK fonts. Deleted the Korean-mixed
template, replaced it with a Hangul-Hanja template.
Updated the Fandol fonts to v0.3. This adds FandolFang which is a FangSong font.
Unfortunately, I discovered that Han uses angle brackets, not angle quotes. Will need
to revise a lot of the CJK fonts for that.
I discovered that my version of NimbusSanL was not recognized as a low-priority alias
for Helvetica by fontconfig. I had an old fork, now replaced by the last version
which now has both names for fontconfig - 'Nimbus Sans L,NimbusSanL'.
In the indexes, removed '‡' markings from monospace fonts :
I had originally intended to delay revising those, but once I became aware of some of
the issues I decided to fix them all before making the site live.
Minor text changes.