- dev-luatex - NTG Mailing Lists

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

Re: charwidth and fullembed fonts
by luigi scarso 04 Feb '25

04 Feb '25

On Tue, 4 Feb 2025 at 10:49, luca boncompagni <lbonco(a)gmail.com> wrote: > > in attach you can find my use case. This is what I have with evince and mutool bake. -- luigi

1 0

missing displacement when leqno is used in rtl paragraph
by Udicoudco 03 Feb '25

03 Feb '25

Hello, I think that displacement of a the of a math display is missing when \leqno is used in RTL paragraph. I attached a patch. Here is a simple test code (it requires the cmr10 font): \catcode`\{=1\catcode`\}=2\catcode`\$=3 \directlua{tex.enableprimitives("",{"pardirection", "textdirection", "Umathquad"})} \tracingoutput=1\tracingonline=1 \showboxbreadth 16384 \showboxdepth 16384 \font\tenrm=cmr10 \textfont0=\tenrm \hsize=345pt\vsize=550pt\parfillskip=0.0pt plus 1.0fil \parindent=15pt\parskip=0pt plus1pt \Umathquad\textstyle=10pt \pardirection=1\textdirection=1 $$1 \leqno(1)$$ \end Udi

5 15

charwidth and fullembed fonts
by luca boncompagni 03 Feb '25

03 Feb '25

Hi to all. I think that if I do a fullembed of a font, it is not correct to include the only charwidth used in the pdf. Think about a font used in a /Form. I tested the attached patch and it seems ok in my case, but I'm not sure it is fully OK. Thanks, Luca

3 2

beforedisplaypenalty has wrong subtype
by Udicoudco 02 Feb '25

02 Feb '25

This is probably just a typo. Since penalty node subtypes are only informational this is a really minor issue. There's a patch attached. Udi

2 1

unexplained difference between luatex and luahbtex
by Petr Olsak 21 Jan '25

21 Jan '25

Hello I am testing new luahbtex and comparing it with luatex because I am considering to set OpTeX with luahbtex engine instead luatex in TeXlive 2025. I have found a difference between lautex and luahbtex behavior which I am unable to explain why. The difference is shown only when huge number of fonts are downloaded by luaotfload and only with selected fonts from internet saved to ~/texmf/fonts. And it manifests only as a little delay (with luahbtex) during dlownloading the fonts to PDF at the end of the document visible if you are watching at the terminal. The PDF result is equal, but I am worried that there is a hidden bug in luahbtex because the difference is unexpected. I teseted this with OpTeX macros, they never use mode=harf. So I expect the same behavior with both engines. The example for testing is somewhat more complicated and it is exactly described at https://tex.stackexchange.com/questions/735071/ The main question is: why luatex behaves correctly (with no delay during downolading fonts) but luahbtex presents a delay in given cases? Where is the difference? Can the luatex developers explain this? Thank you Petr Olsak

3 6

\shapemode does not reset at end of paragraphs
by Udicoudco 19 Jan '25

19 Jan '25

The documentation states that \shapemode should reset to 0 at the end of a paragraph if it is positive, but the following shows otherwise \catcode`\{=1\catcode`\}=2 \directlua{tex.enableprimitives('',tex.extraprimitives ())} \shapemode=1 \showthe\shapemode \quitvmode\par \showthe\shapemode \end I think the fix is a one liner. I attached a patch. Udi

2 3

unexpected delimiter construction with classic tfm cmex font
by David Carlisle 11 Jan '25

11 Jan '25

There are some images of the output and discussion of the \showlists trace in the stackexchange post at https://tex.stackexchange.com/a/734734/1090 but this plain luatex file \tracingonline1 \showboxdepth10 \immediate\write500{\meaning\Arrowvert} $\Arrowvert + \Big\Arrowvert\showlists $ \expandafter\show\the\textfont3 \bye produces the expected || character with pdftex but a { character with luatex even though both claim to be using the same extension font (cmex10) David

3 5