Hi,
just saw "require vignettes to declare their encoding (trunk at
57560)". Having played with non-ascii vignettes (well, a lot of Chinese...)
on-and-off for almost two weeks, I noticed it was a bit odd that a *commented*
\usepackage[utf8]{inputenc} had any effort at all on R's Sweave behavior -
while it is understandable (probably no check is made whether a "%" is
before \usepackage), it is probably wrong.
Besides commented statements having an effect, I found parsing \usepackage for
encoding a bit disturbing - because at one point I was considering adding
non-ascii's to one of the \usepackage options (hyperref can takes non-ascii
for options), and it is possible, or may be even necessary, for
\usepackage's to be in a particular order. There is also another drawback:
if a latex document is multi-language, it is quite customary to have different
encoding in different parts - one reason is to switch fonts and the availability
of glyphs.
FWIW, I noticed it because I was adding/removing \usepackage{}'s and
switching between CJK and XeTeX a fair bit to see which gives me better results.
They have different font selection mechanisms and different
compatible/conflicted TeX packages; CJK on TexLive uses postscript Type 1 fonts
while XeTeX uses Truetype/Opentype - the typeface is the same (both using a
Arphic derivative) but the visual result is different.
HTH,
Hin-Tak
