(Thanks to Eric Norige for noticing this. Somehow I overlooked it....)

Use multiple threads in the 'HashTable' test driver.

(TODO: Add O_DIRECT support for raw streams from regular files.)
(Could've sworn that I had already started that task previously....)

…ded.

Clean up code: tighten down a conditional compilation, remove a stray OMP directive.
Improve performance: precalculate MAX_COUNT and MAX_BIGCOUNT and store with HashTable object.

Improve efficiency of DNA string validator.

…orn-specific RSS limits

…eep-reads2.

'sweep-reads2' loads input reads into a hashbits table and then "sweeps" reads
out of another file if they have a median kmer count of 1, i.e. if half or more
of their k-mers are present in the hashbits table.  Since the median k-mer
count function would have been identical in both the Hashbits and CountingHash
subclasses of Hashtable, I promoted it to a base class function.

One use of this is is to use partitions from diginormed data to recover
the full undiginormed members of the partitions from the original data.
It can also be used as an approximate way to pull out reads from one
data set that overlap with reads in another data set, although cumulative
diginorm + partitioning is a better way to do that if you have the memory.

…generated makefile is missing.

(Bug: Threads hang if the code is not built with internal tracing enabled. Need to fix.)

…urrently

…rate file.

Change 'maxErrorRegion' in Jason's aligner to be an unsigned int.

…ing with the sources.

…ager segments.

…ib-lab#66

where d any positive number

fix query format to be as inout format