* Mike Stay <metaweta@gmail.com> [Feb 06. 2012 16:44]:
There's a pdf reader built into the Chrome web browser because Adobe reader has so many security bugs that we don't allow people to install it at Google.
[...]
That is quite lovely to hear... 8-) Not all "bugs" are bugs: I once wondered why acroread would hang at startup. As it happened there was a stale NFS mount and the thing stalled on scanning the disk starting from / (yes, the root dir). I got an idea what the outbound traffic by acroread might have contained... Here is a document to see how bad things can be: http://www.archive.org/details/combinatoryanaly02macmuoft choose the pdf (17MB). Flipping a page costs 10 secs with xpdf (which is not slow), and still 3 secs with mupdf (which is the Linux equivalent of sumatra-pdf as I found whith checking the sources). This is with a 3 GHz 64-bit CPU, so we are at >=10^10 cycles for rendering one bloody page of a text document. Insane! ...then get the djvu of the same document and rejoice, the performance is almost comparable with what one gets with a dvi file (and the xdvi viewer), which is > keyboard repetition rate which I have at 30 pages per sec. Pretty sure it would do 100/sec. dvi is also the _one_ format where I can point and click in the rendered document to start an editor which opens the corresponding TeX file and jump to the correct line in that file. Point and click in the TeX file lets the dvi viewer jump to the right place in the dvi and draw a neat frame around it (as a hint for the exact place). If you have ever worked with these then everything else, including the whole WYSIWYG nonsense, feels as welcome as a blunt head trauma.