Re: Line breaks and brackets in LaTeX export

[Max Nikulin]

On 19/10/2022 10:57, Ihor Radchenko wrote:
> Juan Manuel Macías writes:
> > Today I have tried with the latest version of tabularray (2022C, the one
> > I tried yesterday was 2022A, included in TeX Live 2022), and the bad
> > results persist. Also, it now returns a compile error when an \empty
> > precedes a \hline. I suspect this package does a pretty drastic
> > redefinition of \\. The [0pt] option still works fine here, though.
>
> Then [0pt] should it be. At least for now, before we have a cleaner
> solution.

It seems when I had a look into latex.ltx first time, I confused which 
branch is executed when length is less than or equal to zero and decided 
that it is the heavier \@xargarraycr. Actually \@yargarraycr do not 
really worry me, so degree of my objection concerning \\[0pt] 
significantly decreased.

\def\@argarraycr[#1]{%
  \ifnum0=`{\fi}${}\ifdim #1>\z@ \@xargarraycr{#1}\else
   \@yargarraycr{#1}\fi}

\def\@xargarraycr#1{\@tempdima #1\advance\@tempdima \dp \@arstrutbox
   \vrule \@height\z@ \@depth\@tempdima \@width\z@ \cr}
\def\@yargarraycr#1{\cr\noalign{\vskip #1}}

I have realized that

| / | <                                | > |
|   | a                                | b |
|   | @@latex:\noalign{\vskip 1em}@@ c | d |

is not a workaround to increase local interval between rows. It may 
cause disrupted vertical rules. Another recipe should be used:

| / | < |                                 > |
|   | a | b @@latex:\rule[-1em]{0pt}{1em}@@ |
|   | c | d                                 |

I believe that a more convenient way to override [0pt] to some other 
length for particular row should exist, but I have no idea which syntax 
should be used.

As to tabulararray, I still consider it as an experimental package. 
Perhaps I will install a more modern container. I am curious what code 
handles \\[0pt]. Likely I should read docs to get impression related to 
design goals and approaches to implement them. The bug tracker of the 
project looks like an appropriate place to ask a question concerning \\ 
variant safe for dumb exporters.

For a while I have the following question. Is \\{} have the same effect 
on tabularray parser as \\\empty?