Re: [Groff] understanding groff's source code

[Ralph Corderoy]

Hi Ulrich,

> In case of questions, should one ask on the list

I'd say ask here;  answers would be part of the public record.  These
answers are all based on a quick look;  I don't know they're correct.

> However, not so for extract().

    node *node_list::extract()
    {
      node *temp = head;
      head = tail = 0;
      return temp;
    }

The list is extracted, not a node;  a pointer to the first element, if
any, is returned.  The caller can walk the list themselves.  The
node_list forgets about those elements.

> Are memory leaks generated?

After extract(), that's the responsbility of the extractor who now has
the list.

> Why is delete_node_list() not called?

Because we haven't finished with the node_list, we just want to extract
what we've appended so far, and then may well come around again to
append more nodes.

> why is node_list in input.cpp but delete_node_list() in node.cpp?

Don't know.

> what does a node-object represent?

A thing representing something in input?  Browse src/roff/troff/node.h.

    $ grep '^class\>.*\<public node\>' src/roff/troff/node.h |
    > awk '{print $2}' | fmt
    line_start_node space_node diverted_space_node diverted_copy_file_node
    extra_size_node vertical_size_node hmotion_node vmotion_node
    hline_node vline_node dummy_node transparent_dummy_node zero_width_node
    left_italic_corrected_node overstrike_node bracket_node special_node
    suppress_node tag_node draw_node
    $

Cheers, Ralph.