[XeTeX] xdv2pdf (was Re: embolden bugs)

[Bruno Voisin]

Le 18 mars 08 à 12:19, William Adams a écrit :

> Filesizes come out smaller using xetex for some .tifs as well (if
> memory serves, Bruno wrote up a lengthy description of one such
> instance).

Your memory serves you right!

There was one instance in which I needed to include 10 huge PDF files  
in a pdfTeX document. By huge I mean each PDF file was about 8 MB,  
obtained from an EPS file of about 40 MB, as produced by Plot3D in  
Mathematica 5 based on data on a 600 x 600 grid. Specifically, there  
were 6 of the files included in one page of the document, and the  
remaining 4 files on another page (ultimately pp. 300 and 302 of <http://dx.doi.org/10.1017/S0022112006004095 
 >).

Proceeding as above, viewing the two pages on screen was a pain (each  
page took minutes to display, and had to be redrawn from scratch  
whenever the page was scrolled by the tiniest amount) and printing  
them was problematic (generally causing crashes after 20 minutes or so  
of processing in a LaserJet).

I ended up converting the PDF files to 600 dpi TIFF bitmaps (which  
were the format and resolution suggested by the publisher for grey  
scale bitmaps), and using XeTeX + xdv2pdf to include the TIFF files in  
my document.

Alas, that's not the end of the story: there were problems at the time  
with XeTeX's typesetting of maths (IIRC the glue before and after  
equations didn't behave the same as with pdfTeX), which eventually  
made me switch back to pdfTeX. I used Apple's Preview to convert the  
TIFF files to PDF files (of about 1 MB each), and used these new  
files. I remember at one time experimenting with tiff2pdf from  
libtiff, but finding out the output of Preview was nicer or smaller,  
I'm not sure.

As usually happens, this was essentially time wasted, given the  
publisher did not use these carefully prepared bitmaps and eventually  
prepared their own low-resolution bitmaps from the original PDF files,  
as can seen in the final published paper.

> I'm fine w/ the maintenance until there's a showstopping reason to
> discontinue it.

I'm rather pessimistic regarding such strategies: maintaining several  
alternative products generally results in the default becoming the  
only one to survive after some time, either because of developer's  
choice or users' choice. Remember for example in the late 90s Apple's  
plan of developing Rhapsody and OS 9 side-by-side, which eventually  
resulted, if I'm not mistaken, into Rhapsody becoming OS X and OS 9  
becoming Classic then dying.

Bruno