15:03:48 <gundalow> #startmeeting Ansible Core Meeting 15:03:48 <zodbot> Meeting started Thu Feb 2 15:03:48 2017 UTC. The chair is gundalow. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:03:48 <zodbot> Useful Commands: #action #agreed #halp #info #idea #link #topic. 15:03:48 <zodbot> The meeting name has been set to 'ansible_core_meeting' 15:04:27 <gundalow> #chair bcoca jtanner mattclay ryansb 15:04:27 <zodbot> Current chairs: bcoca gundalow jtanner mattclay ryansb 15:05:16 <also_alikins> bloop 15:06:06 <gundalow> #chair also_alikins 15:06:06 <zodbot> Current chairs: also_alikins bcoca gundalow jtanner mattclay ryansb 15:06:23 <ryansb> good morning, ansiblites 15:06:27 <gundalow> #info Ansibot is down at the moment, not sure when it will be back 15:06:41 <gundalow> Greetings from Paris 15:06:57 <gundalow> Dates for AnsibleFest London have been announced 15:07:35 <abadger1999> bonjour! 15:07:39 <gundalow> jtyr: over to you 15:07:48 <jtyr> https://github.com/ansible/ansible/pull/19283\ 15:07:54 <jtyr> https://github.com/ansible/ansible/pull/19283 15:08:14 <gundalow> #topic etc_hosts module #19283 15:08:24 <jtyr> I have added the IP verification and some more functionality how to search for the right records ... 15:08:51 <jtyr> now it should definitely have better value over the lineinfile or template ... ;o) 15:09:53 <gundalow> also_alikins: bcoca I believe you two have been reviewing this 15:10:44 <bcoca> still has several open review issues 15:11:01 <abadger1999> sivel's comment hasn't been addressed yet. 15:11:14 <bcoca> nor yours 15:11:16 <gundalow> can someone please take over now please 15:11:28 <also_alikins> I have no serious objections 15:12:14 <bcoca> imports still in wrong spot 15:12:18 * jtanner finally sits down at desk 15:12:42 <jtyr> bcoca: imports should be on the top of the file 15:12:55 <jtyr> any variables should be after imports 15:12:58 <bcoca> jtyr: you are violating pep8 15:13:03 <jtyr> nop 15:13:11 <jtyr> you are by moving it under the variables 15:13:19 <bcoca> yes .. you are ignoring project style 15:13:42 <jtyr> there was no project style documented by the time of creation of the module 15:13:44 <bcoca> which pep8 clearly states is MORE imporant than the details in pep8 15:13:58 <jtyr> the project style was created to make my mode unmergable ... ;o) 15:14:39 <bcoca> yes .. we live to do style changes to make youre modules incompatible </sarcasm> 15:14:43 <abadger1999> It was not. It was created because it makes sense. 15:15:05 <jtyr> bcoca: you don't have to read imports all the time ... you put them in once and then you never look at them again ... so the can stay at the top of the file as where they should be by good programming practice ... 15:15:39 <jtyr> abadger1999: it doesn't make sense ... it's the same like includes in C must be on the top of the file as well ... 15:15:53 <bcoca> jtyr: bring it up in meeting about style guide, the style is what it is, this meeting is not to discuss the style guide 15:15:54 <abadger1999> jtyr: But you don't have then at the top of hte file either. 15:16:07 <abadger1999> you have them at line 22. 15:16:17 <jtyr> comments don't count as code ... 15:16:24 <abadger1999> Exactly. 15:16:37 <jtyr> first comments, then imports then anything else ... 15:16:51 <abadger1999> And for the purposes of hte module, ANSIBLE_METADATA, DOCUMENTATION, RETURNNS, and EXAMPLES are comments. 15:16:52 <jtyr> my module is exactly according to Python standard 15:17:09 <jtyr> abadger1999: no, they are not comments ... they are variables ... 15:17:40 <abadger1999> "For the purposes of" 15:17:47 <bcoca> again, not for discussion now, this is a mute point 15:17:55 <jtanner> "moot" point 15:17:57 <levis_> I think the reality is here that we aren't changing that guideline. So you are fighting a battle that you won't win. So I suggest following direction or closing your PR 15:18:01 <jtyr> if you turn the variables into comments, then there is no problem to move the imports below .. but you decided to make them as variables and therefor the imports must be above them ... 15:18:08 <bcoca> jtanner: no, as in im going to mute both of them .... 15:18:15 <jtanner> very well 15:18:49 <jtyr> so any other comments than imports? 15:19:00 <abadger1999> jtyr: I gave one in the ticket. 15:19:11 <abadger1999> easyfix 15:19:17 <jtyr> abadger1999: object is the default inheritance, no? 15:19:25 <abadger1999> jtyr: no. Only in python3. 15:19:40 <abadger1999> jtyr: in python2, inheriting from object explicitly makes it a new-style class. 15:19:48 <abadger1999> Otherwise it's an old-style class. 15:20:10 <jtyr> eh? ... what's old-style class? 15:20:26 <abadger1999> jtyr: I can't even remember... Probably best to just google it. 15:20:28 <abadger1999> :-) 15:20:54 <levis_> The irony here is not lost on me 15:20:58 <abadger1999> :-) 15:21:16 <abadger1999> Anyhow... I don't think there's anything else we can do with this module via this meeting... 15:21:32 <abadger1999> We've all reviewed it... it's the best it can be bt it doesn't comply with project coding style. 15:21:33 <jtyr> abadger1999: LOL ... http://stackoverflow.com/questions/4906014/python-default-inheritance#comment5462586_4906014 15:21:51 <jtyr> "in most cases, it doesn't matter" ... ;o) 15:23:00 <abadger1999> ah, properties is a big one. 15:23:24 <abadger1999> anyhow... so I guess we take this off the agenda and close the PR? 15:23:34 <abadger1999> you can push the module to galaxy in a role? 15:23:55 <jtyr> merge & close, please ;o) 15:24:22 <abadger1999> Or we can discuss style guidelines some more later in the meeting. 15:25:13 <abadger1999> In either case, I think we should move on from this item. 15:26:33 <albertom> speaking of style.. first import python modules then custome ones (ansible) 15:26:34 * albertom hides 15:27:01 <akasivel> One step at a time there 15:27:08 <bcoca> ^ dont know if we ever said that officially, but that is what i've always follwooed 15:27:57 <abadger1999> <nod> Isn't that one in pep8? 15:28:23 <abadger1999> stdlib<blankline>3rd party<blank line>project-specific 15:28:39 <akasivel> Yes also not enforced by the pep8 tool. There is a flake8-import-order plugin that does it 15:28:54 <akasivel> Although* not enforced 15:29:13 <abadger1999> <nod> 15:29:18 <abadger1999> Okay moving on: 15:29:38 <albertom> thats why i like go 15:29:39 <abadger1999> #topic Fix for mysql user: https://github.com/ansible/ansible/pull/19016 15:30:02 <abadger1999> Looks like jmainguy was asking for other opinions at the end of hte comments. 15:31:01 <Jmainguy> im here now 15:33:37 <abadger1999> Jmainguy: So it looks like you just needed to know if others thought that a warning in the documentaiton if an incorrect privilege string is given was sufficient. 15:34:22 <abadger1999> I'd prefer it to error but I suppose documentation could be sufficient for now.. 15:34:44 <Jmainguy> yeah I would prefer an error, but I am pretty lazy, and the OP doesnt seem to want to do it either 15:35:18 <Jmainguy> I wasnt even aware mysql supported privs per column, so thats pretty cool 15:35:20 <abadger1999> hehe :-) 15:35:27 <Jmainguy> adding that in is gonna be good for folk 15:35:52 <Jmainguy> ok I will respond and ask OP to add that that warning into the documentation 15:36:01 <Jmainguy> then thumbs up it when he/she does 15:36:02 <abadger1999> yeah. 15:36:11 <Jmainguy> ty ty 15:36:32 <abadger1999> I take it that the PR makes the situation better than it currently is any how so I think incremental improvement is good. 15:36:40 <Jmainguy> yup, exactly 15:36:56 <Jmainguy> atm ansible wont work with per column privs, and this PR fixes that 15:37:03 <abadger1999> #action Suggestion to add documentation is fine... incrementatl improvment works for us 15:38:14 <abadger1999> #topic https://github.com/ansible/ansible/pull/19500 Changing assemble to allow alternate sources of files 15:38:46 <abadger1999> So I think this one we discussed pros and cons for a while last time. 15:38:57 <abadger1999> At some point we just have to make a decision. 15:39:06 <bcoca> -1 15:41:52 <abadger1999> bcoca: does template includes work for the examples given in the PR? https://github.com/ansible/ansible/pull/19500/files#diff-fc33cf0f7621e5463eaef7993046ec2eR124 15:42:19 <abadger1999> If so, I guess I don't see a reason to duplicate functionality either. 15:43:24 <bcoca> if not, i can easly add one 15:44:34 <abadger1999> Okay, -1 from me as well. 15:46:05 <abadger1999> also_alikins, ryansb, akasivel, any other committers: If you want to vote on accepting or rejecting "Changing assemble to allow alternate sources of files", now's the time 15:46:23 <jtyr> bcoca: how can you use template to achieve the same like in the example? 15:47:16 <jtyr> I quite like the src_files feature ... not that I would ever use it but if somebody does then it might be nice to have it in Ansible ... 15:47:23 * abadger1999 waits to see results of this 15:47:27 <ryansb> I'm ambivalent, so abstain 15:47:39 <also_alikins> +1 from me 15:48:15 <jtyr> +1 from em as well ... it ads more possibilities how to use the module ... 15:49:22 <bcoca> {% for myfile in include_files %} {% include myfile %}{%endfor%} 15:49:24 <jtanner> i abstain, novote 15:49:31 <jtyr> ... like to use one task to get list of files I would like to concatenate and then pass the list to the assemble module which will do it ... 15:49:59 <jtyr> bcoca: ok ... but you have to create template instead of using just a module ... 15:50:01 <bcoca> i even find the assemble module as overlapping with tempalte 15:50:47 <jtyr> bcoca: almost every other module can be replaced by shell and template modules ... ;o) 15:51:44 <jtyr> bcoca: the point of other modules is to make user's life easier ... and this src_files is one of the features which can make the life easier ... 15:52:38 <bcoca> template yes, shell no, shell is there as a fallback 15:52:58 <bcoca> jtyr: at one point adding too many modules that overlap with trivial options makes things harder 15:53:27 <bcoca> i prefer to use roles that encapsulate these things vs module proliferation 15:53:41 <bcoca> roles are easier for user to customize vs module code 15:53:56 <bcoca> and with include_role we dont have the previous limitations of ordering and reuse 15:54:34 <abadger1999> Okay, current vote is +1:1, 0:2, -1:2.I'd like to have a more solid yay or nay than that. I think I'll record the current vote and reasoning in the ticket. We'll look for more votes next week when people are back from AnsibleAutoMates 15:55:18 <jtyr> bcoca: but you fragment the way how people use ansible ... if there is one module in the core, all people will use it ... if there is 1000 roles, which one will you use? 15:55:31 <abadger1999> #info current votes Accept:1, Reject:2, Abstain:2. Will wait for next week to see how other committers vote on the issue. 15:55:32 <jtyr> I rather use modules than roles ... 15:56:08 <abadger1999> #action abadger1999 to record current votes and rationale in the community ticket for continuing next week. 15:56:44 <jtyr> abadger1999: there was two +1s ... 15:56:50 <abadger1999> jtyr: I think bcoca's thrust is that if there's two modules in the core, half will use one and half will use the other. 15:56:57 <jtyr> one from me and one from also_alikins 15:56:58 <abadger1999> jtyr: Are you a committer? 15:57:20 <abadger1999> I apologize if I missed that. 15:57:35 <jtyr> ah ... I don't count then ... ;o) ... I should do some quick PEP8 commit on that module to get the right to vote ... ;o) 15:57:40 <abadger1999> :-) 15:57:53 <bcoca> maintainer != commiter 15:58:02 <bcoca> all commiters are maintainers, but not all maintainers are commiters 15:58:20 <bcoca> ^ we were trying to use bot to enable them to commit but ... someone reported bot 15:58:47 <jtanner> easiest way to define "committer" is those who can press the green merge button 15:59:23 <jtanner> until there's some other form of rbac, but that would probably have to be through the bot, but ... 16:00:50 <abadger1999> bcoca: https://github.com/ansible/ansible/issues/20440 <=== This is on the agenda and not crossed off but ticket is closed. Does it need discussion or is it done? 16:03:50 <jtyr> jtanner: I would like to be able to press the merge button ... I got couple of PRs still open which would be worth having in Ansible by default ... ;o) 16:04:28 <jtyr> jtanner: If I would have that power, Ansible would become PEP8 compliant over the night ... ;o))) 16:04:31 <jtanner> there's very few people who can do that, who aren't employed by ansible 16:05:14 <bcoca> used to be more .. but we hired them .... 16:06:21 <abadger1999> We need to grow that number. But, to make sure all cards are on the table, people who do pep8 changes without consensus would not likely get in. 16:07:00 <abadger1999> bcoca: do you need us to discuss 20440? or is it done? 16:07:52 <bcoca> nope, user seemed happy with my explanation 16:08:01 <abadger1999> k 16:08:18 <abadger1999> podlesh's item we'll talk about on tuesday. 16:08:21 <bcoca> ah, but not ticket to discuss 16:08:24 <bcoca> was about inheritance 16:08:25 <jtyr> abadger1999: you either follow PEP8 or you don't ;o) ... if all code would be PEP8 compliant it would be significantly better than if it's not compliant ... 16:08:32 <bcoca> what is inherited and what should be 16:08:41 <abadger1999> jtyr: well... the answer to that is. we don't. 16:08:55 <bcoca> jtyr: i disagree, but not going to discuss that now 16:10:12 <abadger1999> #topic Make fileglob work with globbed path components: https://github.com/ansible/ansible/pull/19297 16:10:17 <bcoca> abadger1999: but lets wait till after fest, need more quorum 16:11:56 <abadger1999> IIRC, this was briefly discussed at a previous meeting but no one wrote up results/next actions. 16:12:11 <abadger1999> Anyone care to fill me in on that? (I wasn't here for that meeting) 16:12:44 * bcoca draws blank 16:13:20 <abadger1999> Alright.. I'll crawl the meeting logs to find out the status and we can discuss it next time. 16:13:41 <bcoca> ^ i need to update the 'inheritance' issue since ticket was closed 16:13:46 <abadger1999> #topic Open Floor 16:13:57 <abadger1999> bcoca: Cool :-) That would be great. 16:14:06 <bcoca> new ticket for feb? 16:14:26 <abadger1999> yeah, we need one. this one is getting cluttered. 16:15:26 <abadger1999> bcoca: Go ahead and update the inheritance item in this ticket. I'll go through the items and migrate all the ones that are not crossed off after the meeting. 16:15:57 <abadger1999> #topic Default value of Poll: https://github.com/ansible/ansible/issues/20869 16:16:08 <abadger1999> Here's one that I found yesterday. 16:16:14 <abadger1999> async: + poll: 16:16:22 <abadger1999> In 1.x the default value of poll was 10 16:16:26 <abadger1999> In 2.x it's 0. 16:16:41 <abadger1999> (None, but in the code, None equates to 0 in this case) 16:16:58 <bcoca> abadger1999: ah, i fixed that in my async ticket 16:17:08 <bcoca> set poll to default to 10, otherwise many async cases break 16:17:18 <abadger1999> bcoca: okay, cool. 16:17:26 <bcoca> i'll make note in changelog 16:17:45 <abadger1999> If you want to point that ticket to the async fix and close it that would be great. 16:18:25 <bcoca> done 16:18:38 <abadger1999> bcoca: He also has a feature idea for async: https://github.com/ansible/ansible/issues/20877 Not sure if that's something that would be controversial or if we should just mark it as needs_contributor. 16:18:43 <abadger1999> Thanks 16:22:30 <abadger1999> #topic Open floor 16:22:34 <abadger1999> Okay, anything else? 16:23:34 <abadger1999> If not I'll close in 60s 16:29:10 <ryansb> #endmeeting