Ensure pipe tables

[halloleo]

When using the pantable filter I try to generate pipe tables (because they are supported by other markdown converters like multimarkdown).

This doesn't work normally: All tables are generated as grid tables. However by accident I found out id does work when I set width to a list of 0s. Example:

``` {.table}
---
alignment: DDD
header: true
markdown: true
width: [0, 0, 0]
---
First row,defaulted to be header row,can be disabled
1,cell,It can be
2,more,"It can be  very, very, very, very, very, very, very, very, very, very, very looooong cell."
```

Some questions about this:

• Is this an indented feature?
• Is it documented and i have overlooked it?
• Is it just a lucky coincidence I should not rely on?

[halloleo]

Note: The command line I use is

pandoc -F pantable  input.md -t markdown-simple_tables+pipe_tables 

[ickc]

It's happening on the pandoc side.

pantable parses the CSV into pandoc's native AST, and pandoc decided what to do about it for different outout, including markdown.

In pandoc. there's 4 table syntaxes, where only the grid table supports everything the internal AST support. So if it figures out a certain feature cannot be represented as pipe table, then it will uses grid table for example. And one such feature is the column width, I think.

Another way to force pandoc to output pipe table is to turn off the other three table extensions, such as pandoc -t markdown-grid_table-.... But this has a side effect: if pandoc decide a table cannot be rendered as pipe table, it will choose the next avaible format to output, in this case if other 3 extensions is turned off. the only remaining one will be HTML table (since HTML is valid markdown.)