View large PDF files on Linux
When we make UML plots for our software project, the page size of the generated PDF file can be extremely huge, e.g. larger than 1 metre, due to the complexity in work flow or data structure. If we examine such a PDF file in Okular, even a tiny mouse scrolling will cause the page to reload, during which black blocks will flash around geometric entities.
Since disabling the “Enhance thin lines” function (in Settings → Configure Backends → PDF) of the PDF backend in Okular does not help, I have to resort to the classical but more lightweight PDF reader Xpdf. After testing it with my 1507 x 963 mm UML sequence graph, it can be confirmed that Xpdf performs much better and faster than Okular. When rasterizing a PDF page, Xpdf splits it into several blocks and caches a number of them. When we scroll across cached and uncached blocks, even though Xpdf needs a little time to refresh, there is no flashing black blocks anymore.
In addition to this good performance under extreme working conditions, Xpdf provides a bunch of commands which can be bound to user defined keys. For example, the following configuration in my ~/.xpdfrc enables accelerated scrolling, which is especially suited for viewing large PDF pages.
bind shift-left any scrollLeft(32)
bind shift-right any scrollRight(32)
bind shift-up any scrollUp(32)
bind shift-down any scrollDown(32)
The Postscript and PDF viewer GNU gv performs even better than Xpdf. After loading a large PDF file, there is no refreshing in gv when we scroll the page.
Lessons learned
Some classical Linux software can be relied upon due to their better speed, extensibility and capability for extreme cases. Good examples are Xpdf, Gnuplot, less, etc.