fontlist: support WOFF and WOFF2 fonts by benoit-pierre · Pull Request #14455 · koreader/koreader

benoit-pierre · 2025-10-13T18:38:24Z

This change is

benoit-pierre · 2025-10-13T18:39:05Z

TODO:

base bump with freetype2: fix WOFF support and add WOFF2 support koreader-base#2191 before merging

Commodore64user · 2025-10-13T19:16:00Z

Who let the ~~dogs~~ fonts out?
Woff? woff? Woff?

I'll see myself out.

mergen3107 · 2025-10-13T19:22:04Z

You beat me to it :D

benoit-pierre · 2025-10-13T19:34:07Z

We regularly see this failure with the macOS ARM64 build:

Summary of Failures:

43/73 koreader-testrunner:front / persist                 FAIL             0.60s   killed by signal 11 SIGSEGV

Ok:                72  
Fail:              1

Can someone reproduce locally?

Frenzie · 2025-10-13T20:45:23Z

I'll try if the build finishes within the next 15 minutes or so.

Frenzie · 2025-10-13T20:46:56Z

Bad news on the compilation front, some error about _LIBCPP_ENABLE_ASSERTIONS has been removed, please use _LIBCPP_HARDENING_MODE instead. I'll try a full clean.

Frenzie · 2025-10-13T20:50:24Z

No, it said it again despite starting from scratch.

This is because either HarfBuzz or our usage of HarfBuzz includes -D_LIBCPP_ENABLE_ASSERTIONS=1.

Frenzie · 2025-10-13T20:53:03Z

      | [harfbuzz   6%] Compiling C++ object src/libharfbuzz.0.dylib.p/hb-draw.cc.o
FAILED: src/libharfbuzz.0.dylib.p/hb-draw.cc.o 
env clang++ -Isrc/libharfbuzz.0.dylib.p -Isrc -I../source/src -I. -I../source -I/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/staging/include/freetype2 -I/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/staging/include -fdiagnostics-color=always -D_LIBCPP_ENABLE_ASSERTIONS=1 -Wall -Winvalid-pch -std=c++11 -fno-exceptions -O0 -g -fno-exceptions -fno-rtti -fno-threadsafe-statics -fvisibility-inlines-hidden -DHAVE_CONFIG_H -Og -g -pipe -fno-omit-frame-pointer -idirafter /usr/local/include -ffile-prefix-map=/Users/xxx/src/koreader/base/= -Wno-ignored-optimization-argument -Wno-unknown-warning-option -Wno-array-bounds -Wno-unused-command-line-argument -Wno-self-assign -Wno-parentheses-equality -Wno-macro-redefined -std=gnu++17 -mcpu=apple-m1 -fPIC -MD -MQ src/libharfbuzz.0.dylib.p/hb-draw.cc.o -MF src/libharfbuzz.0.dylib.p/hb-draw.cc.o.d -o src/libharfbuzz.0.dylib.p/hb-draw.cc.o -c ../source/src/hb-draw.cc
In file included from ../source/src/hb-draw.cc:25:
In file included from ../source/src/hb.hh:191:
In file included from ../source/src/hb.h:31:
In file included from ../source/src/hb-blob.h:34:
In file included from ../source/src/hb-common.h:66:
In file included from /Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/c++/v1/inttypes.h:241:
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk/usr/include/c++/v1/__config:49:6: error: "_LIBCPP_ENABLE_ASSERTIONS has been removed, please use _LIBCPP_HARDENING_MODE instead"
   49 | #    error "_LIBCPP_ENABLE_ASSERTIONS has been removed, please use _LIBCPP_HARDENING_MODE instead"
      |      ^

benoit-pierre · 2025-10-13T20:55:42Z

Weird, where is that flag coming from? Anything if you grep the build/…/staging or build/…/thirdparty directories?

Frenzie · 2025-10-13T20:58:18Z

It looks like it's a combination of a mildly older Meson and a newer macOS issue. Updating to 1.9.1 (from 1.7.0) seems to have done the trick, assuming I'm not speaking too quickly.

apache/arrow#46690

benoit-pierre · 2025-10-13T20:58:20Z

I assume that's with KODEBUG=1 (that could explain the difference with the CI), can you check with KODEBUG=?

Frenzie · 2025-10-13T21:08:26Z

Can someone reproduce locally?

I ran this for a few minutes:

while ./kodev test front; do echo yay; done

I was about to say it doesn't look like it's doing any crashing, but then suddenly:

42/73 koreader-testrunner:front / persist                         FAIL             0.30s   killed by signal 11 SIGSEGV
>>> UBSAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1:print_stacktrace=1 LSAN_OPTIONS=exitcode=0 MALLOC_PERTURB_=74 MESON_TEST_ITERATION=1 KO_HOME=/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/spec/run/front_persist MSAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1:print_stacktrace=1 ASAN_OPTIONS=halt_on_error=1:abort_on_error=1:print_summary=1 /Users/xxx/src/koreader/koreader-emulator-arm64-apple-darwin25.0.0-debug/koreader/luajit -e 'require "busted.runner" {standalone = false}' /dev/null --output=gtest -Xoutput=--color --run=front spec/front/unit/persist_spec.lua --config-file=spec/config.lua --exclude-tags=notest --helper=spec/helper.lua --loaders=lua --lazy
―――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――― ✀  ――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――――
has monolibtic? no (dlopen(libs/libkoreader-monolibtic.dylib, 0x0005): tried: 'libs/libkoreader-monolibtic.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OSlibs/libkoreader-monolibtic.dylib' (no such file), '/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/libs/libs/libkoreader-monolibtic.dylib' (no such file), '/Users/xxx/src/koreader/base/build/koreader/libs/libs/libkoreader-monolibtic.dylib' (no such file), '/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/libs/libs/libkoreader-monolibtic.dylib' (no such file), '/Users/xxx/src/koreader/base/build/koreader/libs/libs/libkoreader-monolibtic.dylib' (no such file), '/usr/lib/libs/libkoreader-monolibtic.dylib' (no such file, not in dyld cache), 'libs/libkoreader-monolibtic.dylib' (no such file), '/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/libs/libkoreader-monolibtic.dylib' (no such file), '/System/Volumes/Preboot/Cryptexes/OS/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/libs/libkoreader-monolibtic.dylib' (no such file), '/Users/xxx/src/koreader/base/build/arm64-apple-darwin25.0.0-debug/libs/libkoreader-monolibtic.dylib' (no such file))
lib_search_path: libs/?;/opt/homebrew/lib/?
lib_basic_format: lib%s.dylib
lib_version_format: lib%s.%s.dylib

I'll rerun it a few times with a counter to see if there's any type of regularity to it. I wonder if it could be something like too many open files? But I don't know where to check those kinds of things on macOS.

Frenzie · 2025-10-13T21:13:40Z

We're on 27 iterations by now, so I think it's safe to say there's no clear regularity to it.

cnt=0; while ./kodev test front; do ((cnt++)); echo "yay $cnt"; done

Frenzie · 2025-10-13T21:20:28Z

It happened again by the 47th iteration. But trying again this time it was immediate. Let me see if I can get the full log from that Mac somehow.

Frenzie · 2025-10-13T21:25:54Z

testlog.txt

benoit-pierre · 2025-10-13T21:40:29Z

On Linux, you could use something like this:

▸ ./kodev test -w "gdb -ex run -ex bt -ex quit --batch --args" --meson=-i -v persist

On macOS, maybe something like this?

▸ ./kodev test -w "lldb --one-line run --one-line quit --one-line-on-crash bt run --batch --" --meson=-i -v persist

Frenzie · 2025-10-13T21:44:42Z

It says 'run' does not exist. Maybe it needs to be supplied with a command somehow?

Commodore64user · 2025-10-13T21:51:11Z

Brother is ARMs @kodermike might be of some help here, I believe he too has an emulator on Mx machines.

Frenzie · 2025-10-13T21:52:27Z

Anyway, I'm headed off to bed. Maybe one of the actual Mac users will be able to assist quicker, otherwise you might have to exercise some patience until I can test again. ^_^

benoit-pierre · 2025-10-13T21:52:40Z

Better Linux command ./kodev test -w 'gdb --batch -ex run -ex bt -return-child-result -ex quit --args' --meson=-i -v persist.

benoit-pierre · 2025-10-13T21:53:21Z

Yeah, I got the macOS command from a quick look at one macOS LLDB man page. I'm not sure of the syntax.

Frenzie · 2025-10-13T21:56:56Z

PS The Linux commands both work fine. Nothing to report there afaict.

benoit-pierre · 2025-10-13T21:57:10Z

Some stray run in there, this works on Linux:

./kodev test -w 'lldb --batch --one-line run --one-line-on-crash bt --one-line quit --' --meson=-i -v persist

Frenzie · 2025-10-13T21:57:48Z

Annoyingly I just shut down the Mac, but it is midnight so that's a good thing really. ;-)

benoit-pierre · 2025-10-13T21:58:10Z

It will stop on crash however. Can't find a LLDB equivalent for GDB -return-child-result.

benoit-pierre · 2025-10-13T21:59:38Z

I'm dumb: ./kodev test -w 'lldb --batch --one-line run --one-line-on-crash bt --one-line-on-crash quit\ 1 --one-line quit --' --meson=-i -v persist.

Enough to run in batch mode and still trip the testsuite, even if the exit code is not correct.

Commodore64user · 2025-10-13T22:00:39Z

I'm dumb:

welcome to the club

kodermike · 2025-10-13T22:51:04Z

I'm dumb:

welcome to the club

hold my 🍺 ...

Sadly, my Mx went with the layoff. I can run on an x86, but I think you have that base covered.

Frenzie · 2025-10-14T19:03:26Z

The 59th run failed:

Process 10849 launched: '/Users/rambla/src/koreader/koreader-emulator-arm64-apple-darwin25.0.0-debug/koreader/luajit' (arm64)
Process 10849 stopped
* thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x8)
    frame #0: 0x0000000191506c70 libunwind.dylib`unw_set_reg + 340
libunwind.dylib`unw_set_reg:
->  0x191506c70 <+340>: ldr    w8, [x22, #0x8]
    0x191506c74 <+344>: and    w8, w8, #0xffffff
    0x191506c78 <+348>: cmp    w8, #0x2
    0x191506c7c <+352>: b.ne   0x191506cac    ; <+400>
Target 0: (luajit) stopped.
(lldb) bt
* thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x8)
  * frame #0: 0x0000000191506c70 libunwind.dylib`unw_set_reg + 340
    frame #1: 0x0000000100063148 luajit`err_unwind_jit(version=<unavailable>, actions=<unavailable>, uexclass=5500374307216568834, uex=<unavailable>, ctx=<unavailable>) at lj_err.c:541:7 [opt]
    frame #2: 0x0000000191507bd8 libunwind.dylib`unwind_phase2 + 604
    frame #3: 0x0000000191507964 libunwind.dylib`_Unwind_RaiseException + 800
    frame #4: 0x0000000100006770 luajit`err_raise_ext(g=<unavailable>, errcode=<unavailable>) at lj_err.c:758:3 [opt] [inlined]
    frame #5: 0x0000000100006730 luajit`lj_err_throw(L=0x00000001000b4380, errcode=<unavailable>) at lj_err.c:774:3 [opt]
    frame #6: 0x00000001000068bc luajit`lj_err_run(L=<unavailable>) at lj_err.c:899:3 [opt]
    frame #7: 0x0000000100007114 luajit`lj_err_callermsg(L=0x00000001000b4380, msg=<unavailable>) at lj_err.c:1028:3 [opt]
    frame #8: 0x00000001000110e8 luajit`lj_err_callerv(L=0x00000001000b4380, em=<unavailable>) at lj_err.c:1039:3 [opt]
    frame #9: 0x000000010002f9d0 luajit`serialize_put(w="ot serialize 'function's a minimum of 2 arguments, got: 2ected, got nil)) 'str' (a nil value)should fail to serialize functions\U0000001b[0m\nms)\n99\t\U00000004!a(sample 5!d\U00000006\U00000005", sbx=<unavailable>, o=0x0000000104b4b7b0) at lj_serialize.c:332:5 [opt]
    frame #10: 0x000000010002fb5c luajit`serialize_put(w=<unavailable>, sbx=<unavailable>, o=<unavailable>) at lj_serialize.c:271:10 [opt]
    frame #11: 0x0000000100019428 luajit`lj_serialize_encode(L=0x00000001000b4380, o=<unavailable>) at lj_serialize.c:492:7 [opt]
    frame #12: 0x00000001000f6d2c
    frame #13: 0x0000000100016870 luajit`lua_pcall(L=<unavailable>, nargs=<unavailable>, nresults=<unavailable>, errfunc=<unavailable>) at lj_api.c:1151:12 [opt]
    frame #14: 0x00000001000015a4 luajit`docall(L=0x00000001000b4380, narg=<unavailable>, clear=1) at luajit.c:137:12 [opt]
    frame #15: 0x00000001000014c4 luajit`dostring(L=0x00000001000b4380, s="require \"busted.runner\" {standalone = false}", name=<unavailable>) at luajit.c:191:58 [opt]
    frame #16: 0x0000000100000c34 luajit`runargs(L=0x00000001000b4380, argv=0x000000016fdfeb68, argn=3) at luajit.c:473:11 [opt]
    frame #17: 0x0000000100000b64 luajit`pmain(L=0x00000001000b4380) at luajit.c:561:15 [opt]
    frame #18: 0x0000000100003b1c luajit`lj_BC_FUNCC + 44
    frame #19: 0x00000001000168b4 luajit`lua_cpcall(L=<unavailable>, func=<unavailable>, ud=<unavailable>) at lj_api.c:1179:12 [opt]
    frame #20: 0x000000010000078c luajit`main(argc=13, argv=0x000000016fdfeb68) at luajit.c:596:12 [opt]
    frame #21: 0x0000000181875d54 dyld`start + 7184
(lldb) quit 1

log.txt

Frenzie added this to the 2025.10 milestone Oct 13, 2025

Frenzie approved these changes Oct 13, 2025

View reviewed changes

fontlist: support WOFF and WOFF2 fonts

1ce4bb8

benoit-pierre force-pushed the pr/fontlist_woff_and_woff2_support branch from b3dc697 to 1ce4bb8 Compare October 16, 2025 22:38

Frenzie merged commit c7d4dfa into koreader:master Oct 16, 2025
4 checks passed

benoit-pierre deleted the pr/fontlist_woff_and_woff2_support branch October 16, 2025 23:16

benoit-pierre mentioned this pull request Oct 17, 2025

FR: Woff/Woff2 font support? #14449

Closed

0xstillb pushed a commit to 0xstillb/koreader-thai that referenced this pull request May 9, 2026

fontlist: support WOFF and WOFF2 fonts (koreader#14455)

0f4d4bf

Uh oh!

Uh oh!

Conversation

benoit-pierre commented Oct 13, 2025 • edited by Frenzie Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

benoit-pierre commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Commodore64user commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

mergen3107 commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

Commodore64user commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025

Uh oh!

benoit-pierre commented Oct 13, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Commodore64user commented Oct 13, 2025

Uh oh!

kodermike commented Oct 13, 2025

Uh oh!

Frenzie commented Oct 14, 2025

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants

benoit-pierre commented Oct 13, 2025 •

edited by Frenzie

Loading

benoit-pierre commented Oct 13, 2025 •

edited

Loading

Commodore64user commented Oct 13, 2025 •

edited

Loading

Frenzie commented Oct 13, 2025 •

edited

Loading

benoit-pierre commented Oct 13, 2025 •

edited

Loading