-1 for disabling test without removing the code. Current tests give us at least "something works" status for the feature available to users, without these tests, we can smoothly move to "totally unusable" status. Complete removal of MVCC can be resource-consuming, but if we want to disable tests at least we should hide the public MVCC API or totally prohibit MVCC usage. Also, it can't be done in 2.x release due to backward compatibility.
ср, 2 дек. 2020 г. в 17:28, Nikolay Izhikov <nizhi...@apache.org>: > > Yes, it can be done. However, I don't think that we will get an > agreement on that > > Let’s give it a try and see what happens :) > > > > 2 дек. 2020 г., в 17:23, Вячеслав Коптилин <slava.kopti...@gmail.com> > написал(а): > > > >> Can you start the vote? > > Yes, it can be done. However, I don't think that we will get an agreement > > on that (I just recall the previous discussion). > > And so, we will not remove the MVCC code; on the other hand, nobody will > > support it in the future. We already at this point. This is just my > humble > > opinion. > > > >> It's strange turning off here the whole MVCC tests just because > something > > in the master branch was broken when in the second thread > > On one side, it looks weird, I agree. On the other hand, nobody cares > about > > that and wants to fix tests. This is a stalemate, I think. > > > > Thanks, > > S. > > > > ср, 2 дек. 2020 г. в 16:47, Maxim Muzafarov <mmu...@apache.org>: > > > >> Slava, > >> > >> Can you start the vote? > >> > >> It's strange turning off here the whole MVCC tests just because > >> something in the master branch was broken when in the second thread > >> Community decide to continue MVCC support. Let's start the vote and > >> see what happens. > >> > >> On Wed, 2 Dec 2020 at 16:39, Вячеслав Коптилин < > slava.kopti...@gmail.com> > >> wrote: > >>> > >>>> It will be even worse if our users will face NPE or things like that > >> in > >>> the basic MVCC scenarios just because we don’t tests it. > >>> The feature is not production-ready, and I don't think it is used at > all. > >>> Moreover, MVCC Cache 7, 8, 8, MVCC PDS 1, 2, 4 are already broken > >>> (execution timeouts, flaky test, etc) and I haven't seen anyone who > would > >>> like to fix this. > >>> Why should we waste every contributor's time? IMHO, MVCC suites are > >> useless > >>> and everyone just pushes "re-run possible blockers" button and doesn't > >> care > >>> about MVCC tests at all. > >>> > >>> Thanks, > >>> S. > >>> > >>> ср, 2 дек. 2020 г. в 16:01, Nikolay Izhikov <nizhi...@apache.org>: > >>> > >>>>> I think test suites can be disabled even today > >>>> > >>>> I’m -1 to disable tests without complete removal. > >>>> It will be even worse if our users will face NPE or things like that > in > >>>> the basic MVCC scenarios just because we don’t tests it. > >>>> > >>>> > >>>>> 2 дек. 2020 г., в 15:50, Вячеслав Коптилин <slava.kopti...@gmail.com > >>> > >>>> написал(а): > >>>>> > >>>>> Hi Nikolay, > >>>>> > >>>>>> Why do we need feature in the project that not even tested > >> regularly? > >>>>> Fair enough. However, I am not an expert in this area (MVCC and > >> SQL), so > >>>> I > >>>>> cannot say how much effort it will take. > >>>>> I would say that the opinion of the rest of the community is needed > >> here. > >>>>> > >>>>> Anyway, I think test suites can be disabled even today, while the > >> fate of > >>>>> the MVCC feature can be (and should be) discussed separately. > >>>>> What do you think? > >>>>> > >>>>> Thanks, > >>>>> S. > >>>>> > >>>>> ср, 2 дек. 2020 г. в 15:38, Nikolay Izhikov <nizhi...@apache.org>: > >>>>> > >>>>>> Hello, Slava! > >>>>>> > >>>>>> Yes, this topic comes to the top from time to time :) > >>>>>> > >>>>>>> . I just want to save the time required for getting TCBot's visa > >> and TC > >>>>>> resources. > >>>>>> > >>>>>> Why do we need feature in the project that not even tested > >> regularly? > >>>>>> > >>>>>>> 2 дек. 2020 г., в 15:36, Вячеслав Коптилин < > >> slava.kopti...@gmail.com> > >>>>>> написал(а): > >>>>>>> > >>>>>>> Hello Nikolay, > >>>>>>> > >>>>>>>> +1 to vote for complete MVCC removal. > >>>>>>> It has already been discussed here [1] and, unfortunately, I have > >> not > >>>>>> seen > >>>>>>> an agreement on that. > >>>>>>> > >>>>>>> [1] > >>>>>>> > >>>>>> > >>>> > >> > http://apache-ignite-developers.2346864.n4.nabble.com/Mark-MVCC-with-IgniteExperimental-td45669.html > >>>>>>> > >>>>>>> > >>>>>>> Thanks, > >>>>>>> S. > >>>>>>> > >>>>>>> ср, 2 дек. 2020 г. в 13:05, Nikolay Izhikov <nizhi...@apache.org>: > >>>>>>> > >>>>>>>> +1 to vote for complete MVCC removal. > >>>>>>>> > >>>>>>>> MVCC is a great feature but we should implement it as a > >> first-class > >>>>>>>> feature and not «something that pretends to be working» > >>>>>>>> > >>>>>>>>> 2 дек. 2020 г., в 12:53, Maxim Muzafarov <mmu...@apache.org> > >>>>>> написал(а): > >>>>>>>>> > >>>>>>>>> Hello Slava, > >>>>>>>>> > >>>>>>>>> I think we should vote for MVCC termination of support. If the > >> vote > >>>>>>>>> will be successful than remove it from the source code and > >> disable > >>>>>>>>> MVCC suites. > >>>>>>>>> > >>>>>>>>> Only disabling tests from MVCC sounds not good. > >>>>>>>>> > >>>>>>>>> On Wed, 2 Dec 2020 at 12:32, Вячеслав Коптилин < > >>>>>> slava.kopti...@gmail.com> > >>>>>>>> wrote: > >>>>>>>>>> > >>>>>>>>>> Hello Igniters, > >>>>>>>>>> > >>>>>>>>>> It looks like there is no activity related to maintaining or > >>>>>> developing > >>>>>>>> the > >>>>>>>>>> MVCC feature. > >>>>>>>>>> So, I see no reason to waste TeamCity resources. I propose to > >>>> disable > >>>>>>>> the > >>>>>>>>>> corresponding test suites. > >>>>>>>>>> This has already been discussed here as well [1]. > >>>>>>>>>> > >>>>>>>>>> [1] > >>>>>>>>>> > >>>>>>>> > >>>>>> > >>>> > >> > http://apache-ignite-developers.2346864.n4.nabble.com/Mark-MVCC-with-IgniteExperimental-td45669.html > >>>>>>>>>> > >>>>>>>>>> Thanks, > >>>>>>>>>> S. > >>>>>>>> > >>>>>>>> > >>>>>> > >>>>>> > >>>> > >>>> > >> > >
