Hi,
Hartmut Henkel wrote:
>
>
> When you change \hoffset, the whole RL text moves in the DVI case, but
> not in the PDF case.
In dvi mode there is a big difference between the actual dvi location
(dvi_h, moves r2l or l2r depending on text direction) and where tex
thinks the dvi location is (cur_h, always moves e2l). dvi output is
quite convoluted. In pdf mode, output is much more straightforward.
I think that it would be better to listen to \pagerightoffset in
both cases, but that is open for discussion.
> Also there seems to be some inconsistent magic if
> \pagerightoffset=0pt.
In that case, it defaults to 1inch (pdf) or 2inch (dvi). Yes, that is
a mess (see below). I can easily make the pdf do what the dvi does, but
which behaviour is right?
> I have no idea how it should look like.
Me neither. And internally there are some more variables involved:
\pdfhorigin
\pdfvorigin
\pdfbottomoffset
Ok, \pdfbottomoffset can probably be ignored because luatex at the
moment doesn't do top-to-bottom scripts. But what about those origin
things?
Initex defaults:
\pdfpagewidth 0pt
\pdfpageheight 0pt
\pdfbottomoffset 0pt
\pdfrightoffset 0pt
\hoffset 0pt
\voffset 0pt
\pdfhorigin 1in
\pdfvorigin 1in
Here is the current logic (in DVI):
cur_page_width = (\pdfpagewidth > 0
? \pdfpagewidth
: \wd255 +2*(\hoffset + 1in))
cur_right_offset = (\pdfrightoffset != 0
? \pdfrightoffset
: 2in )
cur_h = \hoffset
<dvi_h_loc> = (\pagedir TLT
? cur_h
: cur_page_width-cur_right_offset - cur_h )
And in PDF, it looks like this:
cur_page_width = (\pdfpagewidth > 0
? \pdfpagewidth
: \wd255 +2*(\hoffset + \pdfhorigin))
cur_right_offset = (\pdfrightoffset != 0
? \pdfrightoffset
: 1in )
cur_h = (\pagedir TLT
? \hoffset + \pdfhorigin;
: cur_page_width-cur_right_offset )
<pdf_h_loc> = cur_h
Best wishes,
Taco
_______________________________________________
dev-luatex mailing list
[email protected]
http://www.ntg.nl/mailman/listinfo/dev-luatex
