----- Original Message ----- > From: "Ademar de Souza Reis Jr." <[email protected]> > To: "Marian Krcmarik" <[email protected]> > Cc: "Chris Evich" <[email protected]>, [email protected], "Virt Test > Development Mailing List" > <[email protected]> > Sent: Thursday, November 14, 2013 5:16:26 PM > Subject: Re: [Autotest] RFC: Release management > > On Thu, Nov 14, 2013 at 09:52:59AM -0500, Marian Krcmarik wrote: > > > > > > > On 11/13/2013 01:11 PM, Lucas Meneghel Rodrigues wrote: > > > > TL/DR; > > > > > > So in a nut-shell, from an ordinary dev's perspective (not release > > > management or maintainer) it's kind of like we're just renaming "next" > > > to "master", correct? > > > > > > Documentation-wise then, where would we suggest somebody's initial > > > experience come from, master or latest release? > > > > > > In general I think this is a good idea, however I'd caution against > > > perceiving any urgency with this change. I think it would be healthier > > > to set some future date, advertise it far and wide, maybe include a > > > PR-freeze also, along with "countdown" announcements along the way. > > > > > > Essentially try to limit the "surprise" factor for people who may not > > > keep up with ML or hangouts more than once or twice every few weeks. > > > For some, the project is important, but keeping up with news may be > > > fairly low on their list. Just a suggestion. > > I take the opportunity of this thread and I would like to > > stress out this point made by Chris. As regular user of > > autotest with virt-test and less frequent contributor I do have > > troubles to keep up with all the frequent and often massive > > changes in autotest especially this year (mostly around virt > > area), I understand the changes are driven by an effort to > > simplify usage of autotest for new comers but It naturally > > increases the "surprise" factors for regular but not very > > frequent contributors. > > Simply said I would appreciate lower speed of various changes - > > in core code, API, conventions, repository and branches layout, > > ... I try to push changes made by group I am part of every 2-3 > > months and I always have to deal with various API, > > repos/branches, convention changes, which is possible but takes > > time which I would like to spend on writing tests themselves. > > Hi Marian. > > The high speed of changes is a good sign, at least when we're > talking about code changes. It shows that autotest and virt-test > have active contributors and a healthy community. Just as with > mainstream projects such as the Linux kernel or QEMU, a high rate > of code changes is good, as long as we're moving forward and not > introducing unexpected instability or gratuitous back and forth > changes. > > Nevertheless, I take your point that we should not forget about > these users who expect stability and can't keep up with the > development pace. In this case, what we can offer is stable > releases and clear documentation of the changes. Let's focus on > improving that. Hi Ademar,
I had mostly API, Convention and repositories layout changes in my mind, not really any code changes such as new tests, extended functionality, etc., Maybe I did not express myself clearly so short description of my workflow may help. I take latest "stable" release of autotest/virt-test because I am into creating some new tests only and I do not want to care about framework itself (in terms of development or debugging possible bugs), I am happily developing my tests, Once I am happy with my tests I pull the latest code (That's the only choice) to rebase my patches on, Mostly the time frame is 2-3 months and then I imo too often have to deal with some of the problems like: - my already existing tests are broken meanwhile (Not because of changes in tested functionality or test itself). - layout of repository/branches/administrative changed - Conventions changed - Exceptions/Conflicts in/with my code due to API/Structures changes. I do not argue specific steps proposed by Lucas in this thread I just hijacked the thread to express my troubles as well as I do not expect to avoid all the problems I mentioned, They are just imo too often and too much for not very frequent contributor who wanna use autotest/virt-test for developing tests and basically There are no maintained stable versions of autotest so I was very close to keep my own fork and not following autotest/virt-test upstream anymore. Yes, having standard stable maintained releases with stable API would help me I believe. There were some plans about stable 1.0 release already almost 2 yrs ago:). Marian > > I believe the new release process proposed by Lucas is a good > move in simplifying things and keeping up with expectations for > newcomers. > > Thanks. > - Ademar > > -- > Ademar de Souza Reis Jr. > Red Hat > > ^[:wq! > _______________________________________________ Virt-test-devel mailing list [email protected] https://www.redhat.com/mailman/listinfo/virt-test-devel
