Add support for multiple cron agent runs per hour

[danieldreier]

Use map function from future parser to allow cron to run the agent more than once per hour.

This re-implements the functionality requested in PR #162, being the ability to run the agent multiple times per hour using cron.

I just hacked this together and wanted to get feedback, both from @tampakrap as to whether it meets his needs, and from other puppet-puppet users because this has a hard dependency on future parser and the compile will fail with current parser. Puppet Forge currently doesn't have a way to indicate a future parser dependency, so we'd need to update the readme with a prominent warning about the future parser dependency.

If this looks like a reasonable approach, I'll write some test coverage, update the readme, and update this PR.

[tampakrap]

rename this to $cron_frequency please? (only here, not in puppet::agent::cron)

[tampakrap]

I like it, thanks for working on it

[igalic]

uhm, does this mean it's puppet 4 only code?

[danieldreier]

@igalic correct, future-parser only. I'm no longer running current parser anywhere so I have no personal need for backwards compatibility, but I wanted to get feedback before going further because I realize that not everybody is ready to make the switch.

If requiring future parser isn't feasible for other puppet-puppet users, I think the second-best option would be to simply parameterize puppet::agent to accept a field passed directly to the minute parameter on the cron resource, defaulting to fqdn_rand(60), and leaving a comment / documentation to suggest adding code like this future-parser-only material in the profile that wraps puppet-puppet for people who want to use it.

If future-parser-only code would cause problems for you, I'm totally fine solving this another way -- I just wanted to propose what I see as the most elegant fix first, in case others are fine with future parser.

[igalic]

gosh, i really want to switch.
i wonder how many of the modules i use would cough up blood the moment i do…

[danieldreier]

@thrnio how's that python tool coming you were working on, for testing catalog compiles? Seems like something @igalic could use to test for future parser support in his current infrastructure, if it's at a point where we could put it on github.

[purplexa]

@danieldreier I want to implement catalog comparison support, then I can take some time to pull the code out and clean it up for public consumption. I hope to spend some time later this week on it.