Both distribute their output over 20 directories, and have part of the output.log deleted. We should reduce this to no more than 4 directories, it makes bugs hard to identify
Also, no nssqa file is being created - which is esential to the evaluation througt the qa_stat script. The information in the QA report about NT / 2000 is unreliable.
problems seems to be that test -s means something different on NT, change the line in init.sh to test -d, should not break things for unix
that the nssqa file was not created was a seperate bug - fixed now in header of QA wrapper. Also fixed init.sh to use the correct versiondir (by it was the -s / -d problem that I suspected)