On Thu, Mar 3, 2016 at 1:17 AM, Yakov Zhdanov <yzhda...@apache.org> wrote:
> Roman, this is not about early and frequent releases, but about special > beta releases. > > I agree with Dmitry and Pavel that we do not need such releases, but need > to mark somehow that feature is experimental: > - add notice to javadocs and readmeio docs (as Dmitry suggested) > - add warning output to logs on the first use of API > I like the warnings in the log a lot. To Roman’s point, I also think that we should consider more frequent releases, especially when there is a feature that we want to make available to the community, experimental or not. > --Yakov > > 2016-03-03 11:50 GMT+03:00 Roman Shtykh <rsht...@yahoo.com.invalid>: > > > I like Vladimir's idea.It is particularly useful when we implement > > integrations with other systems. Releasing them early and, if needed, > > oftenly, may attract more users of those systems and give advantages over > > competitors. > > -Roman > > > > > > On Wednesday, March 2, 2016 2:01 AM, Dmitriy Setrakyan < > > dsetrak...@apache.org> wrote: > > > > > > In my opinion, if a certain feature is experimental, we should simply > > mention it in the release notes and/or documentation. I would not create > a > > separate beta release just for a certain feature. > > > > D. > > > > On Tue, Mar 1, 2016 at 6:57 AM, Pavel Tupitsyn <ptupit...@gridgain.com> > > wrote: > > > > > Hi, > > > > > > I don't think that features like LINQ and ODBC need the same approach > > > as IDEA EAP. IntelliJ has EAP because new features may have bugs > > > or usability issues. With LINQ and ODBC our main concern are not bugs, > > > but unsupported use cases that we did not think of. Known use cases > > > are covered with tests. > > > > > > Beta releases may not get enough attention to gather feedback. > > > I think we should release these features right away and gradually add > > > support > > > for missing use cases, if any emerge. We are not going to change API or > > > break compatibility while adding these things in future. > > > > > > Furthermore, LINQ is only a usability feature. Users can always switch > to > > > raw SQL > > > if something can't be expressed in LINQ. > > > > > > On Tue, Mar 1, 2016 at 10:44 AM, Vladimir Ozerov <voze...@gridgain.com > > > > > wrote: > > > > > > > Igniters, > > > > > > > > I want to circulate again an idea of "betas" for particular product > > > > features. > > > > > > > > For now we have several pretty cool features which are to be released > > > soon: > > > > ODBC driver and LINQ in .NET. Features like this include potentially > > > > infinite amount of use cases and of course we cannot test all of > them. > > I > > > > believe things like this should go through extended release cycle so > > that > > > > we have time to get user's feedback before officially announcing > them. > > It > > > > could be betas, early previews, whatever. > > > > > > > > The main idea is that we have a time window to obtain a feedback and > > > > stabilize the feature. > > > > > > > > This is a common practice for more or less big products. E.g. see > > > Hazelcast > > > > python client [1] and Intellij IDEA EAP 16 [2]. > > > > > > > > Thoughts? > > > > > > > > [1] https://pypi.python.org/pypi/hazelcast-python-client > > > > [2] https://confluence.jetbrains.com/display/IDEADEV/IDEA+16+EAP > > > > > > > > Vladimir. > > > > > > > > > > > > > > > >