Most reader applications have zoom functionality that allows operations such as
fit to page width, fit to page size or full screen. Yet another zoom operation
that I think is very useful is "zoom to bounding box". This zoom would show the
text of a document at the largest zoom possible while showing all the text in a
page.

Other information:
I have implemented zoom to bounding box externally for acroread. For pdf files,
I use ghostscript to figure out the bounding box of the document. Since the
bounding box varies across pages, I heuristically find the maximal bounding box
until it does not change for 5-7 pages. Then based on the screen resolution and
the bounding box, I calculate the geometry for acroread so that *only* the
bounding box is visible on screen and then invoke acroread with this geometry. I
can't do this with evince since it doesn't provide the geometry argument. Please
add the X geometry argument to evince or if possible implement the "zoom to
bounding box" since that is what I do manually otherwise while reading
documents. Thanks!