Yes thanks Hans, Luigi and anyone else involved. This reached texlive pretest and the overprinting seems fixed.
David On Thu, 19 Feb 2026 at 05:03, Mikael Sundqvist <[email protected]> wrote: > Den ons 18 feb. 2026 21:13luigi scarso <[email protected]> skrev: > >> >> >> On Sat, 14 Feb 2026 at 01:00, David Carlisle <[email protected]> >> wrote: >> >>> or in plain >>> >>> \input luaotfload.sty >>> >>> \font\lm="[latinmodern-math.otf]:mode=base;script=math;language=dflt; at >>> 10.0pt]" >>> \font\lms="[latinmodern-math.otf]:mode=base;script=math;language=dflt; >>> at 7.0pt]" >>> \font\lmss="[latinmodern-math.otf]:mode=base;script=math;language=dflt; >>> at 5.0pt]" >>> >>> >>> \textfont0=\lm >>> \scriptfont0=\lms >>> \scriptscriptfont0=\lmss >>> >>> \textfont1=\lm >>> \scriptfont1=\lms >>> \scriptscriptfont1=\lmss >>> >>> \textfont2=\lm >>> \scriptfont2=\lms >>> \scriptscriptfont2=\lmss >>> >>> \textfont3=\lm >>> \scriptfont3=\lms >>> \scriptscriptfont3=\lmss >>> >>> \Umathchardef\int 1 3 "222B >>> >>> >>> >>> $$ >>> \int\limits_{hmmmmmm}\int\limits_{hmmmmmm} >>> $$ >>> \bye >>> >>> >>> >> I have just committed Revision: 77944 >> the relevant part is >> Modified: trunk/Build/source/texk/web2c/luatexdir/tex/mlist.c >> =================================================================== >> --- trunk/Build/source/texk/web2c/luatexdir/tex/mlist.c 2026-02-18 >> 16:11:12 UTC (rev 77943) >> +++ trunk/Build/source/texk/web2c/luatexdir/tex/mlist.c 2026-02-18 >> 20:01:50 UTC (rev 77944) >> @@ -3292,10 +3292,10 @@ >> /*tex v is the still empty target */ >> height(v) = height(y); >> depth(v) = depth(y); >> - if (opentype) { >> - width(v) -= delta; >> - // delta = 0; >> - } >> + //if (opentype) { >> + // width(v) -= delta; >> + // // delta = 0; >> + //} >> /*tex >> >> Attach the limits to |y| and adjust |height(v)|, |depth(v)| >> to >> >> Following Hans, it's almost certainly a partial fix, and needs more >> testing. >> >> (and as per TeX Live 2026 Plan: >> 21feb: Code freeze for final build, major bug fixes only.) >> > > Thank you Hans for the quick fix! > > First of all, sorry to push you into this ic mess of ours again, and sorry > that we did not test the changes from March 2025 earlier. Maybe we need a > better setup for testing during the year. > > Since code freeze for TL26 is so soon, we will now speed up the testing. > Our big set of regression tests will likely(hopefully) catch any remaining > problematic cases. > > Also, I hope that these changes do not introduce any problems for you with > context mkiv. Sorry if they do, since I understand that you are more keen > to work on mkxl. > > Again, thanks for spending time again on luatex, that we are so dependent > on. > >>
