Subject: Re: Suggested change to README.html generation: tables for binaries
To: Julian Assange <>
From: Matthew Orgass <>
List: tech-pkg
Date: 11/09/1999 13:16:26
On 10 Nov 1999, Julian Assange wrote:

> Lynx "support" for tables simply lays all cells out *vertically*,
> regardless of how wide the table is. The reason this works for the
> main NetBSD page is because the cell elements there are not structured --
> the table is simply used to gain some layout width.

  No, lynx only lays out tables verticly if there is more then one line in
table cells (as is the case with the main NetBSD page).  Otherwise, lynx
starts each row on a new line, puts a space between cells, and wraps
anything that goes past the end of the line.  Try my table example from
You will see that it does render each row on one line in this case (it
would look a little worse when the data is longer, but still quite

> You're wrong. The problem here is that lynx *does* understand tables,
> just in an incredibly unuseful way for the presentation of structured
> data. lynx's bogus table layout will interfere with the <pre> hack. Even if
> you could adapt the <pre> hack to lynx, you would then find yourself in
> a situation where you face incompatabilities with browsers that are not
> lynx.

  Aah, I see the confusion here:  I didn't mix <pre> and tables, I had two
examples, one formatted with <pre> and one that uses tables (actually,
lynx does do special formatting with tables within <pre>, but this
violates the DTD, so I certainly agree that this would be an egegious

  In the <pre> version I suggested, I formatted everything with spaces, so
there is no use of tables at all:

> <dl> and <ul> combined solve all these problems, and they are
> extremely old and well rendered tags and well suited to the
> hierarchical nature of the data. No egregious hacks needed.

  The problem with this version is that is takes up much more vertical
space.  This would look something like this:




Matthew Orgass