Daan, Yes, I am. Don't know about others but I was never against discussion on ML. I know there cannot be a process in Apache community that does not involve discussion on ML.
I was primarily concerned with the fact that there is no pressure on the community to make an explicit decision. Requiring and empowering RM to make an explicit call in addition to clear guidelines for what qualifies as a blocker works for me. I would like to add that while the # of users affected is definitely a major factor when ascertaining severity of an issue, should we not consider the technical scope and/or use-case of a defect. For example, let's say there is only one user using basic zone setup with VMware in the community but the bug/regression has caused a major failure like "No provisioning of VMs". Would this be considered a release blocker? Regards, Somesh -----Original Message----- From: Daan Hoogland [mailto:daan.hoogl...@gmail.com] Sent: Monday, August 03, 2015 2:43 PM To: dev Subject: Re: Revisit Process for creating Blocker bugs kewl, Are you sure btw, the openoffice page does state that blockers first have to be discussed on the mailing list, which was I think the argument to start this thread. On Mon, Aug 3, 2015 at 8:20 PM, Somesh Naidu <somesh.na...@citrix.com> wrote: > Daan, that sounds perfect to me! > > Regards, > Somesh > > > -----Original Message----- > From: Daan Hoogland [mailto:daan.hoogl...@gmail.com] > Sent: Monday, August 03, 2015 4:59 AM > To: dev > Subject: Re: Revisit Process for creating Blocker bugs > > Raja, Somesh, > > I want to revise my stand on this slightly; If we make a page like the > openoffice on Somesh shared which states in a little less abstract > ways clear categories that define a blocker we can quicken our > discussions on the subject. An RM could then quickly get feedback and > close or lower blockers that were not according to those standards. > The RM does, in those cases not have to be well informed on every > aspect of ACS. > > the list from the OO page, > <l> > it is a regression in main functionality > it is a crash in main functionality > it is a freeze or loop in main functionality > it is a security issue > it is a privacy issue > it is a data loss > it is a build breaker on one or more of the generally supported platforms > it is an important usability issue in Accessibility (A11Y) > it is a legal issue > it is a translation merge issue > </l> > , is on some points to vague to me to be usable. Also I would want to > be more restrictive. We can not deal with blockers on components if no > active community member use them, so the component/functionality part > should include a strict definition. Also the main part should be well > defined. > > The strictness I propose is only for accepting without discussion that > an issue is a blocker. So anyone, RM, reporter or others can of course > always start a discussion that any more or less trivial issue be > regarded as blocker anyway. > > i want to have one remark on the page on blockers if we create it: > "Please keep in mind that stopping a release, for what is a blocker to > one user may block another user that is in dire need for added > functionality and not blocked by the new issue." > > sound reasonable? > > On Sat, Aug 1, 2015 at 9:36 AM, Daan Hoogland <daan.hoogl...@gmail.com> > wrote: > > Thanks Somesh, It is a usefull link. > > > > Now if for instance an installation can not be used because no initial > > zane can be created, this would be a showstopper. But if a release > > does not have certain obscure features (even as regression) we have a > > discussion. Not whether we should fix it. I am totally with you on > > that. It does not block a release and does not render a deployment of > > such a release completely useless. It will be useless for a group of > > users while it may at the same time remove blockers from previous > > releases for other groups of users. > > > > This dilemma I want to address by introducing the difference. I have > > not seen much 'blocker's amongst the blockers that where reported in > > cloudstack. There were some, sure but most were regressions that would > > hinder some users and as we could well decide that these are blockers > > (and I think we should in *most* cases). they block users and not > > necessarily a release. > > > > > > > > > > On Sat, Aug 1, 2015 at 12:32 AM, Somesh Naidu <somesh.na...@citrix.com> > wrote: > >> Daan, > >> > >> I was using the term "blocker" in context of a release and hence > suggesting involvement of RM in getting a closure. > >> > >> In terms of defect categorization, I found this both relevant and > helpful - https://wiki.openoffice.org/wiki/Showstopper. > >> > >> Regards, > >> Somesh > >> > >> > >> -----Original Message----- > >> From: Daan Hoogland [mailto:daan.hoogl...@gmail.com] > >> Sent: Friday, July 31, 2015 5:52 PM > >> To: dev > >> Subject: Re: Revisit Process for creating Blocker bugs > >> > >> Somesh, please see my replies in line; > >> > >> On Fri, Jul 31, 2015 at 10:31 PM, Somesh Naidu <somesh.na...@citrix.com> > wrote: > >>> Daan, > >>> > >>> While I have the same opinion as you that "No one should be able to > block a release on their own". I also agree that the issue should be posted > to the ML for discussion and it is the responsibility of the person who > posted the defect to do so. > >>> > >>> I am more concerned with the process. My concern is specifically > around this comment from Raja "If no one supports the defect/issue, we will > be putting out a release that has showstopper issues." > >>> > >>> I mean for one, there should be a way for someone to flag an issue as > blocker/showstopper and two, ensure that there is an explicit decision > being made on the severity. > >> > >> ad one: you can send a mail saying "in my opinion the issue > >> CLOUDSTACK-### should be considdered a blocker" > >> ad two: we have such a process, we vote by lazy consensus on technical > >> issues on dev@ > >> > >>> > >>> To me it makes more sense to do this the other way round, that is, the > person who found the issue raises the issue based on his understanding of > the severity/impact. The person who is responsible for triaging (which in > this case is the community) shall use their discretion to justify the > severity and if it doesn't substantiate then downgrade/upgrade the same. > >> > >> this leaves teh community open to being taken hostage by a single > >> person or a small group that keeps bombarding us with blockers. I am > >> being paranoia by past experience. > >> > >>> > >>> Isn't this the general engineering practice? > >> > >> Not to my knowledge, not in this case. Of course we can have a > >> discussion about the semantics of 'blocker'. And then a user may be > >> blocked but that is not this case: our release should be blocked is > >> what blocker means to us. For all practical purposes we don't have a > >> severity 'blocks user'. > >> > >>> > >>> In addition, we'd have a guidelines on defect categorization for > reference that can be looked up while raising a defect. > >> > >> that is a very good idea. > >> > >>> > >>> Regards, > >>> Somesh > >>> > >>> > >>> -----Original Message----- > >>> From: Daan Hoogland [mailto:daan.hoogl...@gmail.com] > >>> Sent: Friday, July 31, 2015 2:34 PM > >>> To: dev > >>> Subject: Re: Revisit Process for creating Blocker bugs > >>> > >>> -1 blocker means blocker and blocks a release. No one should be able > >>> to block a release on their own. We should treat the critical category > >>> as a staging area for those issues. > >>> > >>> On Fri, Jul 31, 2015 at 5:51 PM, Somesh Naidu <somesh.na...@citrix.com> > wrote: > >>>> +1 > >>>> > >>>> Categorizing an issue as blocker/showstopper should need some kind of > moderation. One possibility, voting and/or require approval from certain # > of PMCs. Alternately, this could also be left to the discretion of the RM. > >>>> > >>>> Regards, > >>>> Somesh > >>>> > >>>> -----Original Message----- > >>>> From: Raja Pullela [mailto:raja.pull...@citrix.com] > >>>> Sent: Friday, July 31, 2015 11:15 AM > >>>> To: CloudStack Dev > >>>> Subject: Revisit Process for creating Blocker bugs > >>>> > >>>> Hi, > >>>> > >>>> I am requesting to see if we can revisit the process for creating > "blocker" defects. I heard and do understand that someone can create a > blocker defect and may not actively involve in closing it out and it > doesn't help the product. I am not clear if we are doing this at and > around RC time - however it doesn't matter. > >>>> > >>>> IMHO, feel that someone's involvement should not be taken as a reason > for incorrectly categorizing a defect, meaning a blocker defect being > created as a Critical and opening up a discussion to review. If no one > supports the defect/issue, we will be putting out a release that has > showstopper issues. > >>>> > >>>> Please share your thoughts and concerns for or against lifting this > restriction! > >>>> > >>>> Raja > >>> > >>> > >>> > >>> -- > >>> Daan > >> > >> > >> > >> -- > >> Daan > > > > > > > > -- > > Daan > > > > -- > Daan > -- Daan
