[Dev-luatex] Fw: A strange behavior of `\magstep0`

8 Jun 2025

      Hi, LuaTeX team,
I'm writing this email to append the complete steps about how I found the problem mentioned in the forwarded email.
I want to take a number after a macro and use the content after number.
For example, scale the baseline skip and then output the later content:

\begingroup\endlinechar=-1

\newcount\scalecount

\long\gdef\scaleBase{
\afterassignment\scaleBaseWithFactor
\scalecount=
}

\long\gdef\scaleBaseWithFactor#1{{
\divide\baselineskip by 1000
\multiply\baselineskip by \scalecount
#1
}}

\endgroup

\scaleBase1200
{line1\par line2\par line3}
\bye

This will show a much baseline skip between line1 and line2.
I want to take \magstep values as parameter. But \scaleBase\magstep1{line1\par line2\par line3} did not work as expected.
I found this is because \magstep add a \relax at the end to split itself from possible next number:

$ luatex
This is LuaTeX, Version 1.22.0 (TeX Live 2026/dev/Arch Linux)
 restricted system commands enabled.
**\relax

*\show\magstep
...
\magstep=macro:
#1->\ifcase #1 \@m \or 1200\or 1440\or 1728\or 2074\or 2488\fi \relax .
<*> \show\magstep
?

So I add more 1 macro to strip \relax:

\begingroup\endlinechar=-1

\newcount\scalecount

\long\gdef\scaleBase#1{
%>>>>>>>>>
\ifx#1\magstep
\afterassignment\scaleBaseWithmagstep
\else
%<<<<<<<<<
\afterassignment\scaleBaseWithFactor
\fi%<<<<<<<<<
\scalecount=#1
}

%>>>>>>>>>
\long\gdef\scaleBaseWithmagstep#1\relax#2{
\scaleBaseWithFactor{#2}
}
%<<<<<<<<<

\long\gdef\scaleBaseWithFactor#1{{
\divide\baselineskip by 1000
\multiply\baselineskip by \scalecount
#1
}}

\endgroup

\scaleBase\magstep1%<<<<<<<<<
{line1\par line2\par line3}
\bye

This works fine for \magstep except \magstep0.
\scaleBase\magstep0 make a warning:

(\end occurred inside a group at level 1)

in the log.
In other more complecated cases, it report (\end occurred when \ifcase but not inside a group. So I checked the place sending \magstep.
I checked this by \show#1 in \scaleBaseWithmagstep:

\begingroup\endlinechar=-1

\newcount\scalecount

\long\gdef\scaleBase#1{
\ifx#1\magstep
\afterassignment\scaleBaseWithmagstep
\else
\afterassignment\scaleBaseWithFactor
\fi
\scalecount=#1
}

\long\gdef\scaleBaseWithmagstep#1\relax#2{
\show#1%<<<<<<<<<
\scaleBaseWithFactor{#2}
}

\long\gdef\scaleBaseWithFactor#1{{
\divide\baselineskip by 1000
\multiply\baselineskip by \scalecount
#1
}}

\endgroup

\scaleBase\magstep1
{line1\par line2\par line3}
\bye

And then \scaleBase\magstep1 will \show the definition of \scaleBaseWithFactor, which may mean there is nothing before \relax, just strip \relax as expected. While \scaleBase\magstep0 report:

:!f=tmp sh -c 'luatex $f.tex'
This is LuaTeX, Version 1.22.0 (TeX Live 2026/dev/Arch Linux)
 restricted system commands enabled.
(./tmp.tex
...
\or=\or.
<argument> \or
               1200\or 1440\or 1728\or 2074\or 2488\fi
\scaleBaseWithmagstep #1\relax #2->\show #1
                                           \scaleBaseWithFactor {#2}
l.27 \scaleBase\magstep0{}
apply \or before 1200 to \show.
It seems that:

  *   For \magstep1
     *   \afterassignment insert \scaleBaseWithmagstep after the full \ifcase
  *   For \magstep0
     *   \afterassignment insert \scaleBaseWithmagstep before the 1st \or.

________________________________
From: li lu
Sent: Sunday, June 8, 2025 3:27 PM
To: dev-luatex@ntg.nl 
Subject: A strange behavior of `\magstep0`

Hi LuaTeX Team,
I am writing to report strange behavior of the \magstep0 command , or the original magnification \csname @m\endcsname.
The problem is happened in a macro with \afterassignment and then assign \magstep0 to a count register.
For the LuaTeX version: luatex --version

This is LuaTeX, Version 1.22.0 (TeX Live 2026/dev/Arch Linux)
Development id: 7674
Execute  'luatex --credits'  for credits and version details.

There is NO warranty. Redistribution of this software is covered by
the terms of the GNU General Public License, version 2 or (at your option)
any later version. For more information about these matters, see the file
named COPYING and the LuaTeX source.

The problem is happened in a macro with \afterassignment and then assign \magstep0 to a count register.
A shortest description:
```

\newcount\count
\def\problemhere#1\relax#2{\string#1#2}
\def\todo{
\afterassignment\problemhere
\count=
}

```
For the above macro, \todo\magstep1{HERE} just print HERE while \todo\magstep0{HERE} print \or1200HERE.
It seems that \magstep0 break the \ifcase structure in the \magstep definition, It seems that this is not a expected behavior.
Complete use case steps and more details are on the stackoverflow page

  *   https://tex.stackexchange.com/q/745864/248249

Sincerely,
Yours