Georg Baum wrote:
> - Each loop adds stuff to the preamble
Yes. The preamble stuff has no intelligence at all. It just dumps
output.
> I have more trouble with non lyx-generated documents. The main
> problem is the fact that Parser::tokenize() unconditionally removes
> all whitespace after a command. parse_text() then adds
> unconditionally whitespace at some places. This makes it impossible
> to parse verbatim environments. Another problem is that there are
> commands where whitespace matters, for example after a \hspace{}.
> I started to implement a strategy where Parser::tokenize() does not
> remove any whitespace. This is instead shifted to parse_text(),
> where it is only removed if it is needed. This adds some lines to
> text.C. It is not finished yet, but works already better than the
> original version.
>
> Comments? If nobody has objections, I'll finish this.
André has noted in the past that the current implementation is a hack
that 'worked well enough for him'. I think that you should keep going
with the 'clean' solution that you describe.
Regards,
Angus
