The new version of Adobe Acrobat (which I have the misfortune to be
using at my office) is outputting a warning where we didn't used to get
a warning. Namely, it complains about one particular font in our PDFs.
The warning is:
Cannot extract the embedded font 'LOFCAW+DejaVuSansCondensed.
Some characters may not display or print correctly.
In fact it displays unaccented characters correctly, but leaves a blank
where "accented" characters should appear (à, ö etc.).
This appears to be a problem for this particular DejaVu font, but not
e.g. for CharisSIL. But when I run pdffonts on the output pdf, I don't
see any relevant difference between these two fonts:
ULCJWW+CharisSIL CID TrueType yes yes yes 5 0
...
LOFCAW+DejaVuSansCondensed CID TrueType yes yes yes 809 0
(The information Acrobat gives in the "Fonts" tab of its "Document
Properties" dialog looks similar to the above.) I thought maybe the
problem was with the subsetting of the font (although as you can see,
both the working Charis font and the problematic DejuVu font have only
embedded subsets). IIUC, font embedding is the responsibility of
xdvipdfmx. So I tried forcing this program to embed all fonts would
help, by supplying the -E flag to xdvipdfmx; but this has no effect on
the full embedding of this font, it still embeds only a subset. (Is
there no way to force xdvipdfmx to embed full fonts, rather than
subsets?) And that may not be the problem anyway, since the CharisSIL
font is also a subset.
I've opened these PDFs in Chrome, and they display all the DejaVu
characters just fine. That said, I can't definitively say that the
problem is with Acrobat; for all I know, Chrome is silently substituting
some similar font for the accented characters (although they look right
when I magnify the image--in particular, the 'a' with the grave looks
identical to the 'a' without the grave--so if it's another font, it's an
exceedingly similar one).
Has anyone else run into this problem with the DejaVu fonts and Acrobat?
I can make a MWE if that would help; I'm just hoping someone else has
already run into this problem and has a work-around :-).
Mike Maxwell
University of Washington
--------------------------------------------------
Subscriptions, Archive, and List information, etc.:
http://tug.org/mailman/listinfo/xetex
