Problem
Slurm interprets a -w/--nodelist value containing / as a filename, reading node names from that file (one per line or comma-separated). It also provides -F/--nodefile as an explicit alternative. Spur treats the value as a literal string in all cases.
Slurm also supports node1*2 repetition syntax for --distribution=arbitrary. Spur does not.
Expected behavior
-w /path/to/nodes.txt reads node names from the file
-F/--nodefile accepted as an alias
node1*2 expands to node1,node1 (only relevant with --distribution=arbitrary)
Notes
Both features are rarely used in practice. The file-path detection is the more useful of the two. The repetition syntax is only meaningful with distribution modes Spur doesn't yet implement.
Acceptance criteria
Problem
Slurm interprets a
-w/--nodelistvalue containing/as a filename, reading node names from that file (one per line or comma-separated). It also provides-F/--nodefileas an explicit alternative. Spur treats the value as a literal string in all cases.Slurm also supports
node1*2repetition syntax for--distribution=arbitrary. Spur does not.Expected behavior
-w /path/to/nodes.txtreads node names from the file-F/--nodefileaccepted as an aliasnode1*2expands tonode1,node1(only relevant with--distribution=arbitrary)Notes
Both features are rarely used in practice. The file-path detection is the more useful of the two. The repetition syntax is only meaningful with distribution modes Spur doesn't yet implement.
Acceptance criteria
-w /tmp/nodes.txtreads the file and expands to node names-F /tmp/nodes.txtworks identicallynode1*2expands correctly (if distribution=arbitrary is supported)