merge of my repo

[guilhem]

Hi,

I do my own repo when anyone was available :
https://github.com/Youscribe/sysctl-cookbook

It will be a pleasure to merge it on upstream.

Any comment / hack is welcome.

[nukemberg]

What features are you proposing for merge? merging an entire cookbook is something i will need to think twice about.

[guilhem]

Yes we need to think twice, but it's needed by all of us.
But I hate to see people redo and redo again, because of a lack of will, openness and arrangement.

I understand that you don't wanna break your cookbook, but understand that you are not the only one who use it and we have needs.

Okay, you register first this cookbook @ opscode, but "With great power comes great responsibility".

For the features I want to merge, it's easy, it's all of them.
I add 2 LWRP, some people add centos & redhat support.
We refactor all the code to be cleaner...

Our code is based on yours but with 53 commits of improvements...

[nukemberg]

Please don't take offense. I'm wiling to add features if there is a reason for it and since you know your cookbook much better then myself it makes sense to ask you what extra features it provides.

To the point: it seems I have to review your cookbook before proceeding, but in any case it seems wiser to merge the cookbook feature by feature instead of one-off merge. Are you wiling to prepare these merges for pull or shall I do this on my own?

[guilhem]

Don't worry, I'm not offense :)
I just want to see this going on for the best of all (and the sanity of my repo ^^).

As you want for the pull request, but it seems complicated for me to do this before Monday. If you want it before week-end, do this and I will review, or I will do my best to do this Monday.

Thank you.

[grosser]

👍 plz merge :)

[svanzoest]

Hi,

It seems to me that there too many ways to set a sysctl.
From a lwrp or attributes perspective, I think it would make sense to just take the
https://github.com/opscode-cookbooks/openssh/ approach. One of the issues with a lwrp would be that you could easily set overriding settings in the file.
having it key off of attributes on the node would mean that normal chef precedence would be applied and avoid that problem entirely.

[nukemberg]

I agree wholeheartedly. also consider that rendering from attributes to file then loading this file mean super easy debugging.