perm filename SRCCHK[3,2] blob
sn#363323 filedate 1978-06-27 generic text, type T, neo UTF8
SRCCHK is a different flavor of SRCCOM (SouRCe COMpare, i.e.
line-by-line comparison of text files) based on hashing of full
text lines. It does not ignore past column 80 like SRCCOM does.
On the other hand, it does not attempt to identify the context
of insertions and deletions like SRCCOM does; which is a FEATURE
not a deficiency, it avoids getting all confused. The major
uses of SRCCHK are: (1) after moving a lot of text around in a
file, to see if any text as been lost or duplicated, without
being concerned with where text got moved to; (2) to get a
quick-and-dirty estimate of file changes after SRCCOM went
bezerk due to too many pieces of text moved around or too much
correlation between different pages causing mis-match; (3) in a
file that has lines identical through column 80 but possibly
different beyond, to see the changes, where SRCCOM claims the
lines aren't changed. For further discussion, HELP SRCCOM. See
also BINCOM.
N.b., the hashing of strings (i.e. lines of text) is done by the
perturbed-linear-congruential method, just as in POX. Although
hashing conflicts (two nonequal strings with equal 35-bit hash)
have been constructed artificially, the chance of a hashing
conflict in a natural file is quite low (the hash numbers are
pseudo-random, expected number of conflicts equals 2↑(-35) times
N*(N-1)/2 where N is the number of different lines in the file).