latex-dev: i with macron loops in T1-encoding

[u-fischer]

In latex-dev a definition for i with macron has been added and that loops in T1-encoding

Minimal example showing the bug

\RequirePackage{latexbug}
\documentclass{book}
\usepackage[T1]{fontenc}
%\DeclareTextCompositeCommand{\=}{T1}{\i}{\=\i} %in latex-dev, loops
%\DeclareTextCompositeCommand{\=}{T1}{\i}{something else} %works
\begin{document}
sīc erat scriptum! %similar with \=\i
\end{document}

Log file (required) and possibly PDF file

log-files is a bit long, but basically one gets

\T1\= #1->\@text@composite \T1\= #1\@empty \@text@composite {\add@accent {9}{#1
}}
#1<-\i 

\@text@composite #1#2#3\@text@composite ->\expandafter \@text@composite@x \csna
me \string #1-\string #2\endcsname 
#1<-\T1\= 
#2<-\i 
#3<-\@empty 

\@text@composite@x #1->\ifx #1\relax \expandafter \@secondoftwo \else \expandaf
ter \@firstoftwo \fi #1
#1<-\\T1\=-\i 

\@firstoftwo #1#2->#1
#1<-\\T1\=-\i 
#2<-\add@accent {9}{\i }

\\T1\=-\i ->\=\i 

\=->\T1-cmd \=\T1\= 

\T1-cmd #1->\ifx \protect \@typeset@protect \@inmathwarn #1\else \noexpand #1\e
xpandafter \@gobble \fi 
#1<-\=

\@inmathwarn #1->\ifmmode \@latex@warning {Command \protect #1 invalid in math 
mode}\fi 
#1<-\=

\T1\= #1->\@text@composite \T1\= #1\@empty \@text@composite {\add@accent {9}{#1
}}
#1<-\i 

\@text@composite #1#2#3\@text@composite ->\expandafter \@text@composite@x \csna
me \string #1-\string #2\endcsname 
#1<-\T1\= 
#2<-\i 
#3<-\@empty 

\@text@composite@x #1->\ifx #1\relax \expandafter \@secondoftwo \else \expandaf
ter \@firstoftwo \fi #1
#1<-\\T1\=-\i 

\@firstoftwo #1#2->#1
#1<-\\T1\=-\i 
#2<-\add@accent {9}{\i }

\\T1\=-\i ->\=\i 

\=->\T1-cmd \=\T1\= 

\T1-cmd #1->\ifx \protect \@typeset@protect \@inmathwarn #1\else \noexpand #1\e
xpandafter \@gobble \fi 
#1<-\=

\@inmathwarn #1->\ifmmode \@latex@warning {Command \protect #1 invalid in math 
mode}\fi 
#1<-\=

\T1\= #1->\@text@composite \T1\= #1\@empty \@text@composite {\add@accent {9}{#1
}}
#1<-\i 

[cfr42]

\DeclareTextCompositeCommand{\=}{T1}{\i}{\={\char25}}

[FrankMittelbach]

\DeclareTextCompositeCommand{\=}{T1}{\i}{\={\char25}}

yes, that'll probably do, but why do it in the first place (it works correctly without any declaration afaics)?

I think what was intended was

\DeclareTextCompositeCommand{\=}{T1}{i}{\=\i}

(but I didn't find a reference why the change was made)

[josephwright]

@FrankMittelbach #1453

[josephwright]

Hmm, I think I meant

\DeclareTextCompositeCommand{\=}{T1}{i}{\=\i}

Does that look right?

[FrankMittelbach]

So it is as I thought: the change has a \ too many (the changelog omits the link to that issue)

but then the question is wouldn a similar definition be needed for \=j or is this something that doesn't exist in any spoken or written language?

[FrankMittelbach]

Hmm, I think I meant

\DeclareTextCompositeCommand{\=}{T1}{i}{\=\i}

Does that look right?

yes, and please add the (gh/1453) in the changes logs

[josephwright]

Hmm, I think I meant

\DeclareTextCompositeCommand{\=}{T1}{i}{\=\i}

Does that look right?

yes, and please add the (gh/1453) in the changes logs

OK - I always use the Git blame for changes so to me there is a clear link :)

[car222222]

@FrankMittelbach wrote:
needed for \=j
or is this something that doesn't exist in any spoken or written language?

Very likely not needed, but I could invent a new language that does, if you like!

Do we have any "accented j" characters/glyphs so far?

Check Unicode, maybe! ??

[davidcarlisle]

On Sat, 21 Dec 2024 at 09:19, Chris Rowley ***@***.***> wrote: @FrankMittelbach <https://github.com/FrankMittelbach> wrote: needed for =j or is this something that doesn't exist in any spoken or written language? Very likely not needed, but I could invent a new language that does, if you like! Do we have any "accented j" characters/glyphs so far? Check Unicode, maybe! ??

U+0135 is j with circumflex U+01f0 is j with caron (tuenc.def supports all four of \^j, \^\j, \v{j} and \v\j as composites already) —

…

Reply to this email directly, view it on GitHub <#1587 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAJVYAW4DYCYRNZVWPOXKED2GUXDZAVCNFSM6AAAAABT2X6KNCVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDKNJYGA3DAOBYGE> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[FrankMittelbach]

yes, and please add the (gh/1453) in the changes logs

OK - I always use the Git blame for changes so to me there is a clear link :)

as far as I'm concerned that is a rather lengthly extra step, compared to just go to /issue/number and I'm not sure that blame shows me the releated discussions necessarily (does it? I think I only saw that you made the change but not why)