On Wed, Mar 28, 2018 at 4:28 PM, Robert Eckhardt <reckha...@pivotal.io> wrote:
> > > On Wed, Mar 28, 2018 at 11:20 AM, Dave Page <dp...@pgadmin.org> wrote: > >> >> >> On Wed, Mar 28, 2018 at 2:54 PM, Robert Eckhardt <reckha...@pivotal.io> >> wrote: >> >>> >>> >>> On Wed, Mar 28, 2018 at 4:12 AM, Dave Page <dp...@pgadmin.org> wrote: >>> >>>> >>>> >>>> On Wed, Mar 28, 2018 at 1:37 AM, Robert Eckhardt <reckha...@pivotal.io> >>>> wrote: >>>> >>>>> >>>>> >>>>> On Tue, Mar 27, 2018 at 9:54 AM, Murtuza Zabuawala < >>>>> murtuza.zabuaw...@enterprisedb.com> wrote: >>>>> >>>>>> >>>>>> >>>>>> On Tue, Mar 27, 2018 at 7:06 PM, Robert Eckhardt < >>>>>> reckha...@pivotal.io> wrote: >>>>>> >>>>>>> >>>>>>> >>>>>>> On Tue, Mar 27, 2018 at 6:25 AM, Murtuza Zabuawala < >>>>>>> murtuza.zabuaw...@enterprisedb.com> wrote: >>>>>>> >>>>>>>> On Tue, Mar 27, 2018 at 3:13 PM, Dave Page <dp...@pgadmin.org> >>>>>>>> wrote: >>>>>>>> >>>>>>>>> >>>>>>>>> >>>>>>>>> On Mon, Mar 26, 2018 at 9:26 PM, Robert Eckhardt < >>>>>>>>> reckha...@pivotal.io> wrote: >>>>>>>>> >>>>>>>>>> >>>>>>>>>> >>>>>>>>>> On Mon, Mar 26, 2018 at 2:07 PM, Joao De Almeida Pereira < >>>>>>>>>> jdealmeidapere...@pivotal.io> wrote: >>>>>>>>>> >>>>>>>>>>> Hi Hackers, >>>>>>>>>>> >>>>>>>>>>> @Murtuza: The patch codewise looks good. Nice to see that we are >>>>>>>>>>> using axios instead of jquery ajax calls and that there is some >>>>>>>>>>> coverage >>>>>>>>>>> for the change. >>>>>>>>>>> Nevertheless the Javascript testing looks a bit slim and could >>>>>>>>>>> be improved. Also the DataSorting class could have some other member >>>>>>>>>>> functions like the model validation could be extracted out so that >>>>>>>>>>> it is >>>>>>>>>>> easily tested. >>>>>>>>>>> >>>>>>>>>>> >>>>>>>>>>> @Hackers: This was how we tried to test this feature: >>>>>>>>>>> 1 - Started pgAdmin >>>>>>>>>>> 2 - Opened the query tool for a specific server >>>>>>>>>>> 3 - Executed a SQL statment >>>>>>>>>>> 4 - Pressed the column header to try to order, nothing happened >>>>>>>>>>> 5 - Right clicked the column header to see if it was there the >>>>>>>>>>> option, nothing >>>>>>>>>>> >>>>>>>>>>> This is the behavior that we were expecting, not to have to open >>>>>>>>>>> Data View and then press the icon that is not even near the grid in >>>>>>>>>>> order >>>>>>>>>>> to sort the column. Is this really the way we want people to use >>>>>>>>>>> the grid >>>>>>>>>>> in pgAdmin? Should it be more intuitive? >>>>>>>>>>> >>>>>>>>>> >>>>>>>>>> Have we considered making the grid behave more like excel or >>>>>>>>>> other grids? I think that having the ascending and descending inside >>>>>>>>>> the >>>>>>>>>> column header, we could similarly provide filtering. Something that >>>>>>>>>> would >>>>>>>>>> give users a more intuitive place to look. >>>>>>>>>> >>>>>>>>> >>>>>>>>> Doing the sorting via header clicks is convenient but very >>>>>>>>> restrictive. How do you specify multiple columns to sort by for >>>>>>>>> example? >>>>>>>>> The current design allows you to select columns and the sort order as >>>>>>>>> you >>>>>>>>> see fit. >>>>>>>>> >>>>>>>> >>>>>>> Honestly I'm not sold on my idea, I was just proposing an >>>>>>> alternative in an effort to start a discussion about the user >>>>>>> experience. >>>>>>> Ideally what I'd like to see, maybe this happened, is some user >>>>>>> research. >>>>>>> When we initial worked on refactoring the results grid we made a bunch >>>>>>> of >>>>>>> changes. One of the things we intended to do was to follow up to see how >>>>>>> people were using the grid now so that we could better understand how it >>>>>>> was now being used in order to design and implement features just like >>>>>>> this. Clearly we haven't gotten there yet. >>>>>>> >>>>>>> >>>>>>>> >>>>>>>> Another reason we can't use that because w >>>>>>>> e have already occupied that behaviour for selecting entire column >>>>>>>> when user clicks on header. >>>>>>>> As Dave suggested, I will be merging it with filter dialog meaning >>>>>>>> it will be accessible via direct button on toolbar & keyboard >>>>>>>> shortcut. >>>>>>>> >>>>>>>> >>>>>>> >>>>>>> How are users currently interacting with that filter dialog? >>>>>>> >>>>>> >>>>>> By clicking on the toolbar button as well as keyboard shortcut. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>> >>>>> Sorry I wasn't clear. My question was more along the lines of, do we >>>>> know if people are using the filter functionality? What kind of >>>>> filters are people using? What do they like about it? What do they wish >>>>> they could do above and beyond sorting, etc. >>>>> >>>> >>>> Yes, they are, based on the fact we've had issues reported in the past. >>>> We have no idea how they are using it. >>>> >>>> Sorting is a separate feature that is often requested. The only reason >>>> it's connected here is that both functionalities in pgAdmin 3 were managed >>>> through the same dialogue which based on lack of complaints from users, >>>> generally worked for them. I do know that the proposed "click on headers" >>>> approach will not work for me, as I have multi-part keys in databases which >>>> I like to sort by in specific ways. >>>> >>>> >>> As far as the 'click on header' is concerned I think Murtuza's objection >>> is very valid. I think my overall point is that I believe that sorting and >>> filtering is a legitimate pain but it is a pain I don't fully understand. >>> The solution as presented certainly works it just doesn't feel nice. My >>> concern is that it won't be clear to users what they should do or expect >>> since the UX is unique to pgAdmin. >>> >> >> Right now my concern is getting us back to feature parity in this area >> with pgAdmin 3, as users are complaining. Longer term we can look at >> further improvements or redesigns as resources allow. >> > > Totally fair. > > >> >> >> >>> Less of a concern is that all of these changes are only happening with >>> the 'view data' section of the code and I would assume that people who have >>> queried the DB will also have the need to sort and filter without >>> necessarily rewriting their SQL. >>> >> >> That doesn't concern me - there are already very distinct differences in >> the two modes for the tool. If/when we have a suitable parser on the front >> end we'll be able to dynamically switch between read-write modes and update >> sorting criteria in hand-crafted SQL. Until that large amount of work is >> done, we have two distinct modes, one with generated SQL and writeable >> data, the other with hand-crafted SQL and read-only data. >> > > Also totally fair. As an aside, we are about to spin up some user > interviews for the large number of objects but during those interviews we > were planning to add some follow up questions on the work we did on the > data grid. If pushing this as is until it can be prioritized is something > that is needed to get the heat off of you I'm ok with that. > > I do think we will spend a little time with users so I can better > understand the issues being faced and we will absolutely share that > feedback here. > That will certainly be helpful, thanks. -- Dave Page Blog: http://pgsnake.blogspot.com Twitter: @pgsnake EnterpriseDB UK: http://www.enterprisedb.com The Enterprise PostgreSQL Company
