[Doc/Samples] Rename variables in microbenchmarks YAML samples to avoid conflict

[corrieriluca]

Hi,

I stumbled upon an issue while reading and testing the "Running micro-benchmarks" section of the documentation.

In the example below (from the micro.benchmarks.yml sample), the job variable does not expand as expected (we expect short) at runtime, and thus results in a error from BenchmarkDotNet (it fails parsing its arguments).

  benchmarks:
    source:
      localFolder: .
      project: micro.csproj
    variables:
      filter: "*"
      job: short
    arguments: --job {{job}} --filter {{filter}} --memory
    options:
      benchmarkDotNet: true

Here is the output from crank-agent (shrinked) :

[11:07:27.276]   Arguments:  --inProcess --cli C:\Users\[...]\src\published\micro.exe --join --exporters briefjson markdown --job {
  "DriverVersion": 0,
  "ServerVersion": 4,
  "Id": 0,

  [...]

  "Dependencies": [],
  "CollectDependencies": false
} --filter * --memory

It seems (according to the benchmarks.schema.json file) that job is already used as a global variable by crank to define the job to be executed. That explains why the job variable expands to a big JSON block in the output above.

In this PR, I suggest renaming these variables in the micro.benchmarks.yml and dotnet.benchmarks.yml to filterArg and jobArg in order to correct this issue and avoid any conflict.

Eventually, I replaced the reference to master by main in dotnet.benchmarks.yml for the dotnet/performance repo.

Regards,

[dnfadmin]

All CLA requirements met.

[sebastienros]

Thanks. I indeed introduced the job global property recently.