Oracle obviously has great documentation, and I just realized -hold_jid_ad can involve dependencies on array jobs!
So I use Stata for things. Sometimes I have to (my job, other programs I get), sometimes I like to (survival, LDA, mixed effects models). I also like doing things on clusters, and sometimes in parallel. These don’t always mix, but if you use in your Stata code
local taskid : env SGE_TASK_ID
then you can do (embarrassingly) parallel computing!
with a shell script like, where ZZ is the number of things you want to run:
#$ -t 1-ZZ
stata -b do Stata_Script.do