On Sat, Jul 19, 2008 at 23:51:32 -0400, Celejar wrote: [...]
> I've been printing from IW using the virtual CUPS-PDF printer. I just > tried printing to file (direct to PDF, not to ps), and I now get the > same pdffonts output that you do, nice font names but all 'no's in the > sub column. Ah, so that is the difference: The cups-pdf virtual printer probably uses ghostscript directly, including its "ugly" font naming scheme, while the print-to-file function of iceweasel seems to work via libcairo, resulting in nicer names but wasteful embedding of whole character sets. > OTOH, I find that the file sizes are 10 - 20 percent > *smaller* for pages printed directly to file from IW! I have tried > three different pages: a blog post, a sales invoice and an AP news > article. Hmm, maybe cups-pdf works like more a printer driver, embedding character subsets for each page separately, which could lead to a certain amount of wasteful overlap. I would then expect to see a different name for the same font on different pages, though. Another possible explanation is that libcairo uses a better compression algorithm for the embedded font data, or that the size of the "full" character set of a given font varies with the encoding. > Your test page, OTOH, is about 4 to 5 times *larger* when > printed directly to file. Perhaps the reason is that your page uses > relatively few characters per font, while more typical pages tend to > use much larger subsets, which limits the savings realized by only > including subsets of the fonts in the file? Yes, indeed, that might also explain why the print-to-file function does not bother to figure out the minimum necessary character subset to embed: It does not matter all that much unless somebody puts up a font test page with many different fonts and only a few characters for each individual font. -- Regards, | http://users.icfo.es/Florian.Kulzer Florian | -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]