On Tue, 24 Aug 2010, Hélio Guilherme wrote:
On Mon, Aug 23, 2010 at 11:19 PM, Allin Cottrell
> If you've ever tried to code an RTF table... Sorry, but the
> RTF table syntax is seriously, seriously horrible to the point
> of being almost unusable. (One index of how horrible it is: a
> slightly mal-formed RTF table can easily crash any Microsoft
> Office application. Yes, crash it, not just have the table
> rejected.) I really don't want to go there, but if any other
> brave soul is willing then we'll entertain patches.
I did successfully created a program that outputted a two column
table with header. Maybe is a too simple example, but I would
like to contribute to Gretl code with this. What do you think,
Allin? Should I get my hands dirty on this? ;)
You're welcome, Hélo. One thing to realize is that for printing
model results we need an initial string column (the parameter
name) followed by three numeric columns (coefficient, standard
error, p-value), followed by an additional string column (if
you're interested), namely the number of "stars" or asterisks to
award to the coefficient, based on its statistical significance.
For printing a matrix in RTF, of course, we require an arbitrary
(that is, at compile-time unknown) number of numeric columns.