Place to store todos, future implementation. For more logistic stuff, refer to the README
- Install program with make, i.e. compiled with cc.
- Run repeated steps with make:
- submit job
- build/run docker
- clean "garbage" , or hidden files
- Detailed instruction on what to modify from a template file
- Resource allocation strategy
- Merged error and output file, add a timestamp to it
- combine it with Makefile, so it could be applied by providing CLI arg for script to use, need to provided arg to override name of the job as well
- Remind to submit job only from SCRATCH, if output is to be written, redirect to SCRATCH as well. Then manually transfer it to
PROJECT
- Reproducible flow to that install packages for env, and set it is kernel for jupyter access
- Explanation on
--home and --env XDG_CACHE_HOMEfor exec the shell and submitting the job- Use more plain language
- Optional, turn this to more batched-script styles
- Known problem with conda init , and
eval "conda.bash hook"