Add {attempt} expansion to configuration

[LiterallyUniqueLogin]

Is your feature request related to a problem? Please describe.
I only want to be emailed by SLURM when the job that I submit fails on its last retry.

Describe the solution you'd like
I want to be able to configure this in my profile's config.yaml:

restart-times: 1
cluster:
  sbatch:
     --account XXX
     --partition XXX
     $(if (({attempt} == 2)) ; then echo '--mail-type FAIL --mail-user me@email.com' ; fi)
     # ...

But trying to expand {attempt} throws a NameError.

Describe alternatives you've considered
I used the solution from this issue ( #742 ). I'm creating a new issue, since they're asking for this feature in cluster config, which is deprecated. I want this in profile configuration

What currently works: In my profile's config.yaml:

restart-times: 1
resources:
  - attempt=0 # this doesn't correspond to the retry attempt. That needs to be manually set in the snakefile
cluster:
  sbatch:
     --account XXX
     --partition XXX
     $(if (({resources.attempt} == 2)) ; then echo '--mail-type FAIL --mail-user me@email.com' ; fi)
     # ...

My snakefile:

rule test:
  output:
    'test.txt'
  resources:
    attempt = lambda wildcards, attempt: attempt
  run:
    assert False

This only emails me on the second attempt. But I have to insert the attempt resources into every rule for which I wish this to be true.

[SichongP]

Does setting default-resources work?

[LiterallyUniqueLogin]

@SichongP The current solution with resources works but is hacky, I'm asking for a solution that doesn't involve using resources but can access 'attempts' directly.