- dev-luatex - Mailing Lists

Use-after-free error with \localleftbox?
by davidweichiang＠gmail.com 12 Apr '25

12 Apr '25

Hello, the following MWE produces an error, which looks like it might be a bug: ``` \localrightbox{xyz}% \hfill\break \localleftbox{abc}% \bye ``` Output: ``` This is LuaTeX, Version 1.21.0 (TeX Live 2025/Homebrew) restricted system commands enabled. (./free.tex warning (nodes): attempt to copy free hlist node 399, ignored warning (nodes): attempt to copy free hlist node 369, ignored Underfull \hbox (badness 10000) in paragraph at lines 2--4 [1{/opt/homebrew/Cellar/texlive/20250308_1/share/texmf-dist/fonts/map/pdftex/up dmap/pdftex.map}]) (see the transcript file for additional information)</opt/homebrew/Cellar/texli ve/20250308_1/share/texmf-dist/fonts/type1/public/amsfonts/cm/cmr10.pfb> Output written on free.pdf (1 page, 8930 bytes). Transcript written on free.log. ``` Additionally, the \localleftbox contents do not appear. Additional information from luatex --credits: ``` Compiled with libpng 1.6.47; using 1.6.47 Compiled with lua version 5.3.6 Compiled with mplib version 2.11 Compiled with zlib 1.2.12; using 1.2.12 Development id: 7667 ```

2 1

Font embedding and variable width fonts
by Patrick Gundlach 09 Apr '25

09 Apr '25

Hello all, a somewhat generic question. I am playing with variable fonts, and I can see that I can create font metrics with font:set_variation(...) so the shaper takes the new widths into account. But writing this font into PDF has no visible effect. I assume the subsetter in LuaTeX does not take the variations into account. If this is correct so far: Are there any plans to change this, for example to use the harfbuzz subsetter? Thanks, Patrick

3 6

Unicode data
by Martin Ruckert 23 Mar '25

23 Mar '25

Hi all! I am currently working on moving the hitex engine to utf8, and I discovered that there are some small differences between luatex and xetex (see below). Are these differences on purpose or is there a bug somewhere? Best regards Martin There are differences between xetex and luatex for these five codepoints: codepoint 1: xetex \lccode=13 luatex \lccode=0 codepoint 26, 32, and 39: xetex math family/class/slot = 0/0/2097151 luatex math family/class/slot = 1/0/0 codepoint 65279: xetex \catcode=12 luatex \catcode=9 -- Martin Ruckert Hochschule München Fakultät für Informatik und Mathematik

2 2

Imprecise language in the OpenType Math specification leading to implementation differences amongst math engines
by Apoorv Potnis 19 Mar '25

19 Mar '25

Dear Developers of LuaTeX, I am writing this mail to report a bug in the OpenType Math specification regarding imprecise language, found by Khaled Hosny: [MATH] The algorithm for applying MathKernInfo is incorrect · Issue #1147 · MicrosoftDocs/typography-issues <https://github.com/MicrosoftDocs/typography-issues/issues/1147>. LuaTeX follows the letter of the specification, while XeTeX follows the intent of the specification. This has created compatibility issues for font developers. Thank you. -- Apoorv Potnis

4 4

Re: CMap unrecoverably broken
by Picone, Rico 18 Mar '25

18 Mar '25

Update: Fixed this by updating Ghostscript from 10.04.0 to 10.05.0. —Rico > On Mar 16, 2025, at 12:30 PM, Rico Picone <rico(a)stmartin.edu> wrote: > > I am getting this error now, too. I have the same situation, using LuaTeX to generate a PDF, trying to process that PDF with GS via the command: > > gs -o $@ -dNoOutputFonts -dColorConversionStrategy=/LeaveColorUnchanged -dEncodeColorImages=false -dEncodeGrayImages=false -dEncodeMonoImages=false -sDEVICE=pdfwrite $< > > Denis, did you find a fix? > > —Rico

2 1

italic kern in math missing
by Ulrike Fischer 16 Mar '25

16 Mar '25

With the newest binaries from today (Development id: 7671) I no longer get an italic kern after a f in math with legacy fonts: \tracingoutput1 \showboxbreadth\maxdimen \showboxdepth\maxdimen $f(x)$ \bye ...\hbox(0.0+0.0)x20.0, direction TLT ...\mathon ...\teni f ...\tenrm ( ...\teni x ...\tenrm ) ...\mathoff with pdftex or luatex from texlive 2024 ...\teni f ...\kern1.0764 (italic) ...\tenrm ( ...\teni x ...\tenrm ) With open type fonts / unicode-math there is no problem. Ulrike

2 2

Change in log indentation for luatex 1.21 (texlive 2025 pretest)
by David Carlisle 12 Feb '25

12 Feb '25

The new luatex shows differences in the way it indents and abbreviates log error lines. This is "only" logging but breaks several tests in the latex test suite which is essentially a regression test based around diff-ing normalised logs. If the change is intentional we can update the reference results (and maintainers of other packages using l3build tests will have to do similar) but if this change might be reverted we will hold off updating the test suite. Attached is a small test file and the tl2024 and tl2025 log. The diff is $ diff 24.log 25.log 1c1 < This is LuaTeX, Version 1.18.0 (TeX Live 2024) (format=luatex 2025.1.30) 12 FEB 2025 19:16 --- > This is LuaTeX, Version 1.21.0 (TeX Live 2025) (format=luatex 2025.2.12) 12 FEB 2025 19:16 7c7 < x} --- > x} 10c10 < l.5 ...elax\relax\relax\relax\relax\relax\show\relax --- > l.5 ...ax\relax\relax\relax\relax\relax\show\relax

2 2

Miscellaneous feature addition patches
by Max Chernoff 11 Feb '25

11 Feb '25

Hi Luigi, I've attached some small patches that I've been sitting on for a while (and forgot about until just before the TL25 freeze, sorry). These are all technically feature additions, and I'm aware that LuaTeX is frozen-ish, but the changes are mostly trivial, and there's a proven need for all of these features. I'll copy the commit messages below (and add the patches as attachments): # Add `token.fast_put_next` Currently, `token.put_next` gets the metatable for the passed token userdata object and compares it to the authoritative metatable from the registry. However, this is quite slow; in benchmarking some real-world code, this added 15% to the total runtime of the code: https://github.com/latex3/latex3/pull/1557#issuecomment-2241981665 The new function `token.fast_put_next` skips this error checking, and as such is faster. This does mean that if the passed value is a non-token userdatum, the engine will get confused and crash; however, the `node` and `node.direct` libraries already provide plenty of ways to corrupt the engine if you try hard enough, so this shouldn't be a significant issue. --- manual/luatex-tex.tex | 8 +++- .../texk/web2c/luatexdir/lua/lnewtokenlib.c | 41 +++++++++++++++++++ 2 files changed, 48 insertions(+), 1 deletion(-) # Support setting the environment with "os.spawn". It's not currently possible to create a new process with a custom environment, even though this is fairly useful functionality. This was requested by the l3build developers https://chat.stackexchange.com/transcript/message/65794389#65794389 since right now, l3build uses the shell to do this https://github.com/latex3/l3build/blob/f75745/l3build-file-functions.lua#L1… https://github.com/latex3/l3build/blob/f75745/l3build-file-functions.lua#L1… but this is both non-portable and error-prone. --- manual/luatex-lua.tex | 6 ++++++ source/texk/web2c/luatexdir/lua/loslibext.c | 24 +++++++++++++++++++-- 2 files changed, 28 insertions(+), 2 deletions(-) # Permit usage of the `img` module in `texlua` mode. This is needed for a Lua-based reimplementation of `extractbb` to work without the `ffi` library; see https://github.com/gucci-on-fleek/extractbb/blob/v0.0.7/source/extractbb-sc… for further context. This may also be useful for other scripts; see https://github.com/atticus-sullivan/memoize/pull/2#issuecomment-2613860793 for another example. --- source/texk/web2c/luatexdir/lua/luastuff.c | 4 +--- source/texk/web2c/luatexdir/pdf/pdfgen.c | 2 +- 2 files changed, 2 insertions(+), 4 deletions(-) Thanks, -- Max

2 3

wrong spacing around display with \matheqdirmode=1
by Udicoudco 06 Feb '25

06 Feb '25

This is the current test for deciding not to use shortskip ((d + line_s <= pre_display_size_par) || ((! dir_math_save && l) || (dir_math_save && ! l))) It has two problems: * if eqno_box is null then l (the leqno bool) is always false, so for short unnumbered equations the test is reduced to the value of dir_math_save which is based on \textdirection and \mathdirection, but in this case shortskip should always be used. * dir_math_save is correlated to \textdirection, but the shortskip detection should be based on the position of the equation number, and parfillskip, none of which \textdirection controls. I think \pardirection should be tested here. Here is a faulty example, where shortskip should be used but it is not \matheqdirmode=1\pardirection=1\textdirection=1 Test $$ Test $$ Test \bye In fact if I change \textdirection in this example shortskip is used, which kind of demonstrates the second point. For a better example of the second point you can test \matheqdirmode=1\textdirection=1 Test $$ Test\leqno(1) $$ Test \bye Whereas \textdirection and \mathdirection are opposites dir_math_save is true and leqno bool is false, hence short skp is used but the output clearly shows it should not be, as the equation number is right below the text. Note that I also tested with LuaTeX 1.18.0 to make sure this is not the fault of recent changes. I attached a patch. Udi

2 4

Ignore null paragraph before display equation
by Udicoudco 05 Feb '25

05 Feb '25

Currently LuaTeX ignores "null paragraphs" (see the manual at the end of section 3.3.2 "How it works"), but it is actually doing so only for paragraphs created with end_graf. enter_display_math ignores paragraphs if they only contain local_par nodes, but not if they also contain dir nodes. This is somewhat inconsistent, so it would be nice if enter_display_math would discard paragraphs like end_graf. I attached a patch. It is almost identical to the change in https://gitlab.lisn.upsaclay.fr/texlive/luatex/-/commit/908e51adaad4b4048fe… but it is even smaller, since enter_display_math already discard paragraphs containing only local_par nodes. Even though this change is less disruptive, there might be more concerns about backwards compatibility today than three years ago, if this is the case a new value to \matheqdirmode can be added. Udi

3 11