[chef-dev] Re: Re: [chef] Re: notifies :before


Chronological Thread 
  • From: Lamont Granquist < >
  • To:
  • Subject: [chef-dev] Re: Re: [chef] Re: notifies :before
  • Date: Tue, 26 Nov 2013 11:49:59 -0800

On 11/24/13 7:57 PM, Xabier de Zuazo wrote:
  Thanks for your detailed reply. Don't worry. I know it's a long ticket
  and this makes it really difficult to review :-)

  I reviewed my patches and, in short, the events of the ticket were:

1. My first PR.
2. You recommended me to use why-run to predict updated_by_last_action.
3. Updated to use why-run.
4. You told me that CHEF-3589 broke my path.
5. Rebased and logic changed to work with CHEF-3589.
6. Won't fix.

  Since then, it seems that you have not made ​​big changes to why-run
  logic. I mean, this code seems to work in current master. I rebased it
  here (fixing some small conflicts):

https://github.com/onddo/chef/compare/CHEF-2421-3

  I tested it with some simple recipes and seems to work.

  I know my proposal is far from being perfect. But I find it strange
  that it was rejected without giving me a chance to improve it. Maybe I
  was misguided, but AFAIK nobody wrote me about it :-S

  Anyway, based on this code, do you think it appropriate for me to open
  a new ticket and propose as a solution? (I can not open the previous
  ticket) Or you just don't like my solution and you are looking for
  something else?

  Cheers,

What is your feature supposed to do now? The thing that you wanted to do was to be able to use the old behavior of why-run mode in order to be able to figure out if there were any queue'd up converge_by blocks, but before the blocks were executed, in order to get a kind of "would_modify_by_action?" check. We completely broke the ability to do that, and as its noted in the ticket its difficult to even predict in the general case that a resource will be updated before you call the action. So what are the conditions under which you expect these before notifications to fire? And why can't you restructure your run_list so that those resources simply come first and are idempotent? Why can't you restructure your run_list to use an after notification?

This smells like its become a feature-for-the-sake-of-a-feature where there's other, better, clearer ways to get the job done.




Archive powered by MHonArc 2.6.16.

§