Given a LaTeX file, foo.tex
, that requires xelatex, one can convert it to PDF in the following two ways:
- Run
xelatex foo
(which directly generatesfoo.pdf
) - Run
xelatex -no-pdf foo
(which generatesfoo.xdv
) followed byxdvipdfmx foo
(which converts the xdv file tofoo.pdf
)
Do the two approaches, in effect, produce the same pdf file? Or is there some advantage to (1) (or indeed (2))?
(The context for this is that in the past I had always assumed that with a latex file that does not require xelatex it was preferable to call pdflatex
rather than calling latex
and converting the DVI to PDF. Indeed, in the past latexmk
with the -xelatex
argument would follow route (1), however newer versions follow (2), which initially concerned me, in analogy with my pdflatex comment above. Having found "latexmk with xelatex xdv mode" I suspect that (1) and (2) are essentially equivalent, but I wanted to confirm this. In particular I want to be sure the two routes are equivalent even if one is using packages such as microtype or hyperref.)
Best Answer
The approaches are identical because xetex calls xdvipdfmx unless you specify -no-pdf.
xetex is not like pdftex or luatex that have a modified back end that generates pdf, it always generates (extended) dvi but it runs xdvipdfmx at the end and deletes the intermediate files unless
--no-pdf
option is given.