<br><br>On Tuesday, July 15, 2014, Jay S. Bryant <<a href="mailto:jsbryant@electronicjungle.net">jsbryant@electronicjungle.net</a>> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
John,<br>
<br>
So you have said a few times that the specs are a learning process.<br>
What do you feel with have learned thus far using specs?</blockquote><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"></blockquote><div><br></div>- we are inexperienced at doing design work<div>
- we are inexperienced at considering the full impact of a change</div><div>- agreeing on the problem statement is more important than agreeing a solution</div><div>- it's easy to confuse design work with implementation work, and we need to get better at separating the two</div>
<div>- specs more easily allow cross-project collaboration and non-dev stakeholders to provide early feedback when compared to blueprints or implementation changes<span></span></div><div>- specs don't make it easy to compare two solutions to the same problem, but I hope they will</div>
<div>- blueprints are even more cumbersome than ever</div><div>- specs are terrible at tracking work items and assignees</div><div><div> </div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
I think you<br>
had the hope that it would help organize targeting blueprints and not<br>
missing things for a release.  Do you feel that is working?<br>
<br>
I would like to hear more about how you feel this is working.<br>
<br>
Personally, initially, having specs made sense given that they are a<br>
better way to spark discussion around a design change.  They were<br>
working well early in the release for that purpose.<br>
<br>
Now, however, they have become a point of confusion.  There is the<br>
question of "when is just a BP sufficient" versus when is neither<br>
required.<br>
<br>
I think this is part of the learning process.  Just worth having<br>
discussion so we know for the next round what is needed when.<br>
<br>
Jay<br>
<br>
On Sun, 2014-07-13 at 16:31 -0600, John Griffith wrote:<br>
><br>
><br>
><br>
><br>
> On Sun, Jul 13, 2014 at 8:01 AM, Dolph Mathews<br>
> <<a href="javascript:;" onclick="_e(event, 'cvml', 'dolph.mathews@gmail.com')">dolph.mathews@gmail.com</a>> wrote:<br>
><br>
>         On Wed, Jul 9, 2014 at 1:54 PM, Jay Bryant<br>
>         <<a href="javascript:;" onclick="_e(event, 'cvml', 'jsbryant@electronicjungle.net')">jsbryant@electronicjungle.net</a>> wrote:<br>
>                 I had been under the impression that all BPs we going<br>
>                 to require a spec.  I, however,  was made are in<br>
>                 today's cinder meeting that we are only requiring<br>
>                 specs for changes that change the user's interaction<br>
>                 with the system or are a large change that touches the<br>
>                 broader cinder code base.<br>
><br>
>         That's generally where we use blueprints in Keystone, anyway.<br>
>         If the change has no impact on end users, deployers or other<br>
>         projects, then it doesn't need a spec/blueprint. For example,<br>
>         a refactor only affects developers of Keystone, so I'd argue<br>
>         that blueprints are unnecessary.<br>
><br>
><br>
><br>
>         The premise of a "large change that touches the broader ...<br>
>         code base" requiring a blueprint is flawed though - we don't<br>
>         want to review large changes anyway ;)<br>
><br>
><br>
> ​Just have to say I really like this last point... also even though<br>
> I'm the one who made that statement the problem with it is that it's<br>
> rather subjective.<br>
><br>
><br>
> My position is and continues to be that specs are a learning process,<br>
> we're hammering it out and these conversations on the ML are helpful.​<br>
>                 This seemsto make sense to me.  The user's commit<br>
>                 message and unit tests should show the thought of the<br>
>                 changes impact.<br>
><br>
>                 Jay<br>
><br>
>                 On Jul 9, 2014 7:57 AM, "Dugger, Donald D"<br>
>                 <<a href="javascript:;" onclick="_e(event, 'cvml', 'donald.d.dugger@intel.com')">donald.d.dugger@intel.com</a>> wrote:<br>
>                         Much as I dislike the overhead and the extra<br>
>                         latency involved (now you need to have a<br>
>                         review cycle for the spec plus the review<br>
>                         cycle for the patch itself) I agreed with the<br>
>                         `small features require small specs’.  The<br>
>                         problem is that even a small change can have a<br>
>                         big impact.  Forcing people to create a spec<br>
>                         even for small features means that it’s very<br>
>                         clear that the implications of the feature<br>
>                         have been thought about and addressed.<br>
><br>
><br>
><br>
>                         Note that there is a similar issue with bugs.<br>
>                          I would expect that a patch to fix a bug<br>
>                         would have to have a corresponding bug report.<br>
>                         Just accepting patches with no known<br>
>                         justification seems like the wrong way to go.<br>
><br>
><br>
><br>
>                         --<br>
><br>
>                         Don Dugger<br>
><br>
>                         "Censeo Toto nos in Kansa esse decisse." - D.<br>
>                         Gale<br>
><br>
>                         Ph: 303/443-3786<br>
><br>
><br>
><br>
>                         From: Dolph Mathews<br>
>                         [mailto:<a href="javascript:;" onclick="_e(event, 'cvml', 'dolph.mathews@gmail.com')">dolph.mathews@gmail.com</a>]<br>
>                         Sent: Tuesday, July 1, 2014 11:02 AM<br>
>                         To: OpenStack Development Mailing List (not<br>
>                         for usage questions)<br>
>                         Subject: Re: [openstack-dev] [all][specs]<br>
>                         Please stop doing specs for any changes in<br>
>                         projects<br>
><br>
><br>
><br>
>                         The argument has been made in the past that<br>
>                         small features will require correspondingly<br>
>                         small specs. If there's a counter-argument to<br>
>                         this example (a "small" feature requiring a<br>
>                         relatively large amount of spec effort), I'd<br>
>                         love to have links to both the spec and the<br>
>                         resulting implementation so we can discuss<br>
>                         exactly why the spec was an unnecessary<br>
>                         additional effort.<br>
><br>
>                         On Tue, Jul 1, 2014 at 10:30 AM, Jason<br>
>                         Dunsmore <<a href="javascript:;" onclick="_e(event, 'cvml', 'jason.dunsmore@rackspace.com')">jason.dunsmore@rackspace.com</a>> wrote:<br>
><br>
>                                 On Mon, Jun 30 2014, Joshua Harlow<br>
>                                 wrote:<br>
><br>
>                                 > There is a balance here that needs<br>
>                                 to be worked out and I've seen<br>
>                                 > specs start to turn into<br>
>                                 requirements for every single patch<br>
>                                 (even if<br>
>                                 > the patch is pretty small). I hope<br>
>                                 we can rework the 'balance in the<br>
>                                 > force' to avoid being so strict that<br>
>                                 every little thing requires a<br>
>                                 > spec. This will not end well for us<br>
>                                 as a community.<br>
>                                 ><br>
>                                 > How have others thought the spec<br>
>                                 process has worked out so far? To<br>
>                                 > much overhead, to little…?<br>
>                                 ><br>
>                                 > I personally am of the opinion that<br>
>                                 specs should be used for large<br>
>                                 > topics (defining large is of course<br>
>                                 arbitrary); and I hope we find the<br>
>                                 > right balance to avoid scaring<br>
>                                 everyone away from working with<br>
>                                 > openstack. Maybe all of this is part<br>
>                                 of openstack maturing, I'm not<br>
>                                 > sure, but it'd be great if we could<br>
>                                 have some guidelines around when<br>
>                                 > is a spec needed and when isn't it<br>
>                                 and take it into consideration when<br>
>                                 > requesting a spec that the person<br>
>                                 you have requested may get<br>
>                                 > frustrated and just leave the<br>
>                                 community (and we must not have this<br>
>                                 > happen) if you ask for it without<br>
>                                 explaining why and how clearly.<br>
><br>
>                                 +1 I think specs are too much overhead<br>
>                                 for small features.  A set of<br>
>                                 guidelines about when specs are needed<br>
>                                 would be sufficient.  Leave the<br>
>                                 option about when to submit a design<br>
>                                 vs. when to submit code to the<br>
>                                 contributor.<br>
><br>
>                                 Jason<br>
><br>
><br>
>                                 _______________________________________________<br>
>                                 OpenStack-dev mailing list<br>
>                                 <a href="javascript:;" onclick="_e(event, 'cvml', 'OpenStack-dev@lists.openstack.org')">OpenStack-dev@lists.openstack.org</a><br>
>                                 <a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
><br>
><br>
><br>
><br>
><br>
><br>
>                         _______________________________________________<br>
>                         OpenStack-dev mailing list<br>
>                         <a href="javascript:;" onclick="_e(event, 'cvml', 'OpenStack-dev@lists.openstack.org')">OpenStack-dev@lists.openstack.org</a><br>
>                         <a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
><br>
><br>
>                 _______________________________________________<br>
>                 OpenStack-dev mailing list<br>
>                 <a href="javascript:;" onclick="_e(event, 'cvml', 'OpenStack-dev@lists.openstack.org')">OpenStack-dev@lists.openstack.org</a><br>
>                 <a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
><br>
><br>
><br>
><br>
>         _______________________________________________<br>
>         OpenStack-dev mailing list<br>
>         <a href="javascript:;" onclick="_e(event, 'cvml', 'OpenStack-dev@lists.openstack.org')">OpenStack-dev@lists.openstack.org</a><br>
>         <a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
><br>
><br>
><br>
<br>
<br>
<br>
_______________________________________________<br>
OpenStack-dev mailing list<br>
<a href="javascript:;" onclick="_e(event, 'cvml', 'OpenStack-dev@lists.openstack.org')">OpenStack-dev@lists.openstack.org</a><br>
<a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
</blockquote></div>