I think we've tried to be pretty clear about what we consider to be a "backward compatible" change in the versioning guide that we have provided. Specifically, others should be able to build upon the nifi-api, the rest api, and util packages that are commonly depended on like processor-utils (which contains most of the standard validators), and the site-to-site client.
Personally, I would not expect any skinning to be compatible even between bug releases, because there's no specific api that is provided for it. > On Jan 6, 2016, at 4:16 PM, Tony Kurc <[email protected]> wrote: > > Is there is a possibility of folks having skinned or extended the UI? Would > these changes be expected to work on minor version revisions? > > On Wed, Jan 6, 2016 at 4:14 PM, Simon Ball <[email protected]> wrote: > >> The new UI looks fantastic, and seems to be heading in a very good >> direction. >> >> One thought I have is around the upgrade experience. Given that we will >> likely end up with different sized elements to the existing UI, people may >> find existing flow layouts end up somewhat jumbled with the repositioning >> of elements. This should be easy to solve, with a migration the existing >> attributes to a transformed (proportionally scaled?) coordinate system. >> >> Impact of this may be limited, but for early adopters who’ve put a lot of >> investment into their layouts, it has the potential to end up making a mess >> if we don’t provide some sort of migration. >> >> An alternative would be to limit the design to follow the existing >> bounding box dimensions, but I think this would be a shame, and preclude >> some of the lovely visual improvements that are emerging in the new mockups. >> >> Simon >> >>> On 6 Jan 2016, at 20:29, Mark Payne <[email protected]> wrote: >>> >>> I think we could certainly do that, where we create a separate branch to >> make all those changes. >>> However, I do think that comes with some downsides. >>> >>> The UI codebase would likely be very different. Any change that is made >> to the 0.x baseline would >>> have to be made in two codebases, and if the 1.x branch is very >> different, git rebases and the like may not >>> work very well and trying to keep those changes in sync can quickly >> become a rather large burden and be >>> very error-prone. >>> >>> We've also been talking about revamping the UI since around version >> 0.2.0. It's been in discussions for >>> quite a long time, and it's just taken a while to come to fruition. >>> >>> Additionally, the new design, I believe, will provide a far better user >> experience for low-resolution displays >>> by providing the docked panels instead of letting everything rest at the >> top of the screen, as well as high >>> resolution displays by providing different types of components and >> getting rid of some of those gradients. >>> >>> While I can understand the idea of "This is a major change so it should >> wait until 1.0.0," I also think that since >>> it provides the same functionality and is not breaking in any way, it is >> not necessary to do so. >>> >>> Some key points that I believe we will need for 1.0.0 include >> multi-tenant capabilities as well as HA. Changes to >>> the UI I think could be accomplished quite a bit before those >> capabilities are addressed. So while we could >>> bench the changes on a branch for quite a while for 1.0.0, I don't >> believe it's really necessary and I think that >>> the pros outweigh the cons in this case. >>> >>> So I'm a +1 for introducing these changes in 0.6.0 and not waiting for a >> 1.0.0 release. >>> >>> Thanks >>> -Mark >>> >>> >>>> On Jan 6, 2016, at 3:10 PM, Sean Busbey <[email protected]> wrote: >>>> >>>> Doesn't this mean we could start moving towards 1.0 by keeping these >>>> UI changes isolated in an appropriate branch and make the completion >>>> of multitentant dataflow (and whatever other features) the complete >>>> marker? >>>> >>>> If we need the UI in a distributable form, e.g. so that we can start >>>> gathering feedback, we can do that by having a 1.0.0-alpha or the >>>> like. >>>> >>>> On Wed, Jan 6, 2016 at 1:40 PM, Matt Gilman <[email protected]> >> wrote: >>>>> Tony, >>>>> >>>>> That is correct. The UI redesign is necessary for a 1.0 feature. >>>>> Additionally, the UI changes are limited to styling and positioning of >>>>> certain elements for controlling the dataflow. >>>>> >>>>> Matt >>>>> >>>>> On Wed, Jan 6, 2016 at 2:33 PM, Tony Kurc <[email protected]> wrote: >>>>> >>>>>> Matt, >>>>>> I'm not sure I followed. I think you said we have two things: >>>>>> A) ui redesign >>>>>> B) multi-tenant dataflow >>>>>> >>>>>> We need A before B, and B will be in 1.0, so we need A in a release >> before >>>>>> 1.0. >>>>>> >>>>>> >>>>>> Tony >>>>>> >>>>>> On Wed, Jan 6, 2016 at 2:17 PM, Matt Gilman <[email protected]> >>>>>> wrote: >>>>>> >>>>>>> One of the items that's needed to go over the 1.0 cliff is support >> for >>>>>>> multi-tenant dataflows. Scoping user authorization to portions of a >>>>>>> dataflow will require a major bump due to API changes and the like. >> Part >>>>>> of >>>>>>> the motivation for this UI redesign is laying the foundation for that >>>>>>> effort. Additionally, we'll get the added benefit of supporting a >> more >>>>>>> responsive layout and modernized look and feel in the meantime. >>>>>>> >>>>>>> Matt >>>>>>> >>>>>>> On Wed, Jan 6, 2016 at 1:28 PM, Sean Busbey <[email protected]> >> wrote: >>>>>>> >>>>>>>> +1 to "UI redesign warrants a major version increment" >>>>>>>> >>>>>>>> I know that we're "pre 1.0", but this sounds like it's time to >> figure >>>>>> out >>>>>>>> what we need to include to go over that cliff. >>>>>>>> >>>>>>>> On Wed, Jan 6, 2016 at 12:22 PM, Tony Kurc <[email protected]> >> wrote: >>>>>>>> >>>>>>>>> Looks great so far! >>>>>>>>> >>>>>>>>> I saw the target release of 0.6, which surprised me a bit. I would >>>>>> have >>>>>>>>> expected this significant of a change would warrant a major release >>>>>>>>> increment. >>>>>>>>> >>>>>>>>> On Wed, Jan 6, 2016 at 1:08 PM, Rob Moran <[email protected]> >> wrote: >>>>>>>>> >>>>>>>>>> Greetings NiFi community, >>>>>>>>>> >>>>>>>>>> NIFI-1323 [1] has been mentioned in (at least one) previous >> thread. >>>>>>> In >>>>>>>>> case >>>>>>>>>> you are unaware, it involves beginning work on a series of UI >>>>>>>>> improvements. >>>>>>>>>> >>>>>>>>>> I'd like to point your attention to the wiki page [2] where you >> can >>>>>>>> read >>>>>>>>> up >>>>>>>>>> on factors that are driving this effort. >>>>>>>>>> >>>>>>>>>> [1] https://issues.apache.org/jira/browse/NIFI-1323 >>>>>>>>>> [2] >>>>>>>>>> >>>>>>>> >>>>>> >> https://cwiki.apache.org/confluence/display/NIFI/Redesign+User+Interface >>>>>>>>>> >>>>>>>>>> Thanks, >>>>>>>>>> Rob >>>>>>>>>> >>>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> >>>>>>>> -- >>>>>>>> Sean >>>>>>>> >>>>>>> >>>>>> >>> >>> >> >>
