I am a member of a project that uses CVS for tracking a large software project.  Recently, we encoutered a problem where version 1.2 of a file which had its head version of 1.5 was very slow to extract.  We are running 1.12.9 -- the debian stable release.

I tried various other versions of CVS (1.11.17, 1.11.21, 1.12.11, 1.12.12) and they all exhibited the same behavior.  The case that caused a problem was the file had 300K lines and 200K fragments (separate add/del ops).  This causes the on the order of 2 minutes per file to perform an update.  I have isolated the problem to the rcs code that looks like it is at least order n^2 (actually n*m where n is the # of lines and m is the number of fragments).