Based on PR#43777 by Kamel Derouiche, with some clean up by me.
Version 1.20, 9/3/2008. This version fixes a bug that caused lucy
to fail when there was too much information on the FASTA header lines
in the input files. Lucy had a 256 character buffer for reading
lines from the input files. If any FASTA header lines were longer
than 256 characters, the remaining characters would get read as part
of the FASTA sequence. The solution implemented in this version is
simply to increase the buffer size to 4096 characters. While that
still leaves the potential for the same error to occur with extremely
long header lines, the limitation that this entails seems reasonable,
and this should fix the problem for all pratical purposes.
---------------------------------------------------------------------
Version 1.19, 12/30/2003. This version fixes a bug that could cause
sequences to be rejected incorrectly in the vector detection step
(phase 6). Because of the way that lucy compares "tags" in the
target sequence with "tags" in the vector sequence, some bases in
the target sequence could get counted more than once in the tally
of bases that match the vector sequence. In rare instances, this
could cause the sequence to exceed the minimum threshold for
rejection, as a result of random sequence similarity.
The reporting of the CLB range in the -debug output file has also
been changed. If the CLB range begins with the first base of the
sequence, then the left coordinate of the CLB range will be reported
as 1 (instead of 0). The range "CLB 0 0" still indicates an empty
CLB range.
546d012f69