[Feature Request] option to control width gestimation of tables.

[terefang]

today simple_tables and pipe_tables as well as HTML-Tables are subject to an automatic mechanism gestimating their column widths by various means.

there are various usecases that either want gestimation to be performed always or never.

• [Feature Request] allow simple markdown tables such as pipe_tables also observe column widths #10427
• Pipe tables / HTML: option to disable the column width computation algorithm for long lines #10111

for the sake of continuity also the current behaviour should be preserved most likely as the default.

Describe your proposed improvement and the problem it solves.

the proposal is to have an option called --smart-table-widths that take the following parameters:

• "auto" being the behavior as it is now, should be the default.
• "off" never compute them as for @sphakka case (above).
• "on" always compute them as for my case.(above).

in my "always" case the gestimated widths become available to lua-filters and the typst target to ensure proper layout regardless of table and column sizes.

in @sphakka "never" case the gestimated widths are discarded in favor of the reflow capabilities of the HTML/CSS target.

[jgm]

I believe that only pipe_tables is subject to this. Simple tables never have widths calculated; multiline tables and grid tables always hav them. So this only affects pipe tables.

[jgm]

Accordingly, the option, if there is one, should be --pipe-table-widths=smart|always|never.

[terefang]

the option naming sounds reasonable to me.

[bpj]

see #10427 (comment)

[Mathnerd314]

Not sure if this is actually related, but, as mentioned in the user guide there are "ugly" pipe tables, like:

fruit|price
---|---
apple|2.05
pear|1.37
orange|3.09

As an option I would like to have Pandoc output these "ugly" pipe tables, i.e. skip the width calculation, use three dashes for each column in the table header, and just output the pipe character or maybe pipe plus spaces like apple | 2.05 for each row.

[terefang]

if i see this correctly that would be the --pipe-table-widths=never case

[jgm]

if i see this correctly that would be the --pipe-table-widths=never case.

No. The proposal above was for an option that affects parsing of tables, controlling whether width information is computed. The request above has to do with rendering. It is irrelevant to this issue.

[terefang]

ok ... sorry for my miss-interpretation

[Mathnerd314]

The request above has to do with rendering. It is irrelevant to this issue.

OK I opened #10453.