Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Make time filter more usable #4981

Merged
merged 31 commits into from
Jul 16, 2018

Conversation

betodealmeida
Copy link
Member

This PR addresses #4292. The main change is that it combines the since and until controls into a single one, time_range:

time_range

This has a few benefits:

  • One cannot select a start date that occurs after the end date.
  • Common time ranges, eg, "last week", can be selected with a single click.

Additionally, there are other improvements to the UX:

  • Freeform has now been combined with the calendar selector. Users can type a date, a freeform string ("last Sunday") or use the calendar. Pressing enter closes the dialog.
  • Selecting a common range, eg, "last week", will update both calendars with the corresponding start/end dates. This allows quickly selecting a range and then fine tuning it, or switching from relative to fixed.
  • Choosing a date from the calendar closes the calendar.

(@vylc, note that instead of using "Relative" and "Fixed/Freeform" like in your mocks, I used "Range" and "Start/End". This is because the latter can also be relative, since freeform can take strings like "last Sunday", eg. I'm open to better naming suggestions.)

Tests

  • Opening saved slices and bookmarks with only since and until work.
  • Filter box works.
  • Unit tests were added.

@codecov-io
Copy link

codecov-io commented May 11, 2018

Codecov Report

Merging #4981 into master will decrease coverage by 15.91%.
The diff coverage is 68.46%.

Impacted file tree graph

@@             Coverage Diff             @@
##           master    #4981       +/-   ##
===========================================
- Coverage   77.15%   61.24%   -15.92%     
===========================================
  Files          44      373      +329     
  Lines        8892    23600    +14708     
  Branches        0     2735     +2735     
===========================================
+ Hits         6861    14453     +7592     
- Misses       2031     9132     +7101     
- Partials        0       15       +15
Impacted Files Coverage Δ
superset/assets/src/explore/controls.jsx 47.32% <ø> (ø)
superset/assets/src/explore/visTypes.js 55.55% <ø> (ø)
...set/assets/src/dashboard/deprecated/v1/reducers.js 0% <ø> (ø)
superset/assets/src/visualizations/filter_box.jsx 27.27% <0%> (ø)
superset/viz.py 77.95% <100%> (-3.4%) ⬇️
superset/assets/src/logger.js 95.65% <100%> (ø)
superset/legacy.py 18.3% <100%> (+3.6%) ⬆️
superset/views/core.py 73.11% <100%> (+0.25%) ⬆️
superset/utils.py 89.16% <100%> (+0.95%) ⬆️
.../explore/components/controls/DateFilterControl.jsx 60.55% <52.94%> (ø)
... and 336 more

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 19ac6e1...03462cb. Read the comment docs.

@mistercrunch
Copy link
Member

@GabeLoins , mind taking a look into this as this falls under similar as your recent work?

@mistercrunch
Copy link
Member

For UI consistency with the work on MetricsControl and FiltersControl I'd go with tabs over pills.

@gabe-lyons
Copy link
Contributor

I'll take a look. Plus plus for using react-bootstrap Tabs for switching between Range and Start/End for visual consistency.

We have also been thinking about ways to improve this section on our side. For future work, if you're interested, I think it'd be great to collapse all three components in the Time section into a single pill that lets you configure all this stuff

};


Copy link
Contributor

@gabe-lyons gabe-lyons May 14, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

accidental newline?

@gabe-lyons
Copy link
Contributor

Another thought- for the metric and filter popovers I had the state in the popover only persist when the user clicked Save. I prefer the save functionality so if the user makes a mistake or changes their mind its easy to get back to the original state. For these benefits, consistency, I think it would be an improvement to include a concept of saving to this popover

export default class DateFilterControl extends React.Component {
constructor(props) {
super(props);
const value = props.value || '';
const value = props.value || defaultProps.value;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

isn't this redundant of how react already deals with defaultProps?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I was unsure about this and was afraid of breaking things. I assume it's protecting cases where value is explicitly set to null?

Copy link
Contributor

@gabe-lyons gabe-lyons May 22, 2018

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

PropTypes (if the field is required) will prevent against null being passed in- I'd say drop the || defaultProps.value.

If the field can't be required, I'd still say its not needed- controls across explore take in the value prop at face value and don't do this check. For it to create a problem someone would have to start setting value as null in a higher level explore component which I think would be a bigger issue.

// for start:end(includes freeform)
since: DEFAULT_SINCE,
until: DEFAULT_UNTIL,
isFreeform: {},
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I feel like this value should be a boolean if its prefixed with is?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What if we initialize it as:

isFreeform: { since: false, until: false }

I can also rename it to freeformFields.

@@ -88,6 +88,7 @@
"react-addons-shallow-compare": "^15.4.2",
"react-alert": "^2.3.0",
"react-bootstrap": "^0.31.5",
"react-bootstrap-datetimepicker": "0.0.22",
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

just noticed this component is deprecated, should we switch to the non-deprecated package?

https://github.com/YouCanBookMe/react-datetime

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Let me take a look at how much effort is needed to use that one... I remember testing it when I was looking for calendar components, and it didn't work for some cases but TBH I don't remember the details.

const INVALID_DATE_MESSAGE = 'Invalid date';
const SEPARATOR = ' : ';
const FREEFORM_TOOLTIP = t(
'Superset supports smart date parsing. Strings like `last sunday` or ' +
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

[honest question] is this functionality worth supporting? It seems like if I wasn't a power user I wouldn't know which keywords work and which don't. For example, would "2 weeks ago" work? What about "the quarter before last"

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Even if we decided we're better off without it, it'd be hard to deprecate (would break charts).

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, without feedback informing if the input is valid this doesn't seem very usable. We could do a backend call and show it in red if not supported?

this.state.dttm = value;
this.state.type = 'fix';
if (value.indexOf(SEPARATOR) >= 0) {
this.state.type = 'startend';
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Since put this string in a constant at the top of this file it seems like it'd be best if you referenced it from the constant as well? Otherwise why use a constant?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good point, let me fix this.

const grainOptions = TIME_GRAIN_OPTIONS.map((grain, index) => (
<MenuItem
onSelect={this.setCustomRange.bind(this, 'grain')}
key={'grain-' + index}
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

generally using index as a key is considered an antipattern (https://medium.com/@robinpokorny/index-as-a-key-is-an-anti-pattern-e0349aece318)

also style nit: I don't think you need to add that grain- prefix

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ah, #TIL. Will fix, though in this case it should be safe (no reordering or updates in the list).

key={'grain-' + index}
eventKey={grain}
active={grain === this.state.grain}
>{grain}
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: {grain} should be on its own line

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Will fix.

}
setDatetime(dttm) {
this.setState({ dttm: dttm.format().substring(0, 19) });
updateRefs() {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this seems like it adds quite a bit of indirection. Would it be possible just to pass value as a prop to these components rather than setting their state via a ref? https://github.com/YouCanBookMe/react-datetime#api

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

This is needed for freeform text. The calendar component uses moment to parse the string, and if it fails the state is not set. This prevents us from initializing the component saved with freeform values.

</PopoverSection>
<Tab eventKey={1} title="Range">
<FormGroup>
{timeFrames}
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

maybe this should be its own third tab? It seems like the Range tab is a little busy with the radio buttons and also the dropdown menus

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@vylc, any thoughts here? Personally I like this design, since the common ranges and the dropdown serve the same purpose, and allow for a lot of control in a single place.

@gabe-lyons
Copy link
Contributor

image

@williaster
Copy link
Contributor

There's some great improvements here! My thoughts on other issues being discussed

  • ++ tabs, we should try hard to move to a world of visual consistency.
  • ++ using the condensed padding that @GabeLoins mentioned, again visual consistency and we purposefully used condensed tabs for his work on metrics/filters.
  • I think using Ok/cancel is also a good idea, that's a pretty common web paradigm
  • I think "Range" and "Start/End" are confusing, they are both ranges in the end right? I'm not sure of the best language here, it seems like "range" is really just smart defaults for choosing the start date? will continue to think on this
  • I like having both the radio buttons + dropdown in the same window pane, IMO it doesn't seem useful to have the radio buttons stand on their own.

@michellethomas
Copy link
Contributor

For Start/End it would be nice to have the language make it clear if the range will be inclusive or exclusive. Druid and sqlalchemy data sources do not filter on the date ranges as both inclusive, and it can be especially confusing when looking at sums over a range (like in big number or table viz).

It could be nice if druid/sqlalchemy were consistent, but changing it now would make existing slices incorrect (as people have adjusted to this difference). When we move to using druid sql, this will be consistent.

I'll try to look around more for specific suggestions. Maybe something like Starting after and Up to? It's a little verbose though.

@betodealmeida
Copy link
Member Author

Yeah, I'm unsure about the naming. Maybe we could use "simple" and "custom", like the filters?

@michellethomas, how are they different? Is one of them inclusive and the other exclusive?

@michellethomas
Copy link
Contributor

Yeah after looking into it, it may be difficult to message because it depends on whether you are using a date or not (and also which engine you are using).

For us using dates (not datetime):
druid start date inclusive / end date exclusive
sqla (with presto) start date exclusive / end date inclusive

sqla tries to be inclusive on both ends and will add a filter like WHERE "ds" >= '2018-05-14 00:00:00' AND "ds" <= '2018-05-18 00:00:00' but it’s getting excluded because of the lexicographic order.

Here's a discussion of the issue. I think it was just dropped because we couldn't find a regression, but there's still an inconsistency we should discuss (maybe out of scope for this PR) https://apache-superset.slack.com/archives/C7G8CG0LR/p1512418244000100

@elibrumbaugh
Copy link
Contributor

elibrumbaugh commented May 23, 2018

++ Tabs
++ Condensed padding

I think tabs work great to separate different concepts or types of tasks. As Max pointed out they should be visually consistent. The tabs in the gif look a bit like buttons to me.

I agree with Chris that "Range" and "Start/End" are confusing since they are both types of ranges. I also view the use of "range" in this context as a type of smart default.

A few thoughts:

  • Regardless of what route we take, we should improve the label clarity.
  • Could we separate smart defaults into one tab and the user's ability to pick a custom range into another? This would mean moving the current last option from "range" into the second tab.
  • I think the separation illustrated below would help separate the two types of user desire: 1. give me great defaults and 2. give me the ability to customize them when they don't meet my needs.

For example:

Defaults

  • [Yesterday]
  • [Last week]
  • [Last month]
  • [Last year]

Custom

  • [Last v] [#] [days v]
  • [Start date]
  • [End date]

An open question I have that effects the order of the tabs is what will people use most often? We should order accordingly.

@vylc
Copy link

vylc commented May 23, 2018

Looks great!
(1) I like the organization that @elibrumbaugh suggested. Can we name the tab "Relative" vs. Defaults. Custom is good.
(2) Re: the inclusive/exclusive that @michellethomas brought up, I think that belongs in an info hover as opposed to trying to name the fields accordingly
(3) Ok/Cancel are good. @GabeLoins do you want to adopt this as well for your metric popover (vs. Save/Close)? For consistency, I'd also move your buttons to the lower right since that's more natural than over to the lower left. Cancel (no color) should be to the left of OK (teal superset color)

@betodealmeida
Copy link
Member Author

(1) I like the organization that @elibrumbaugh suggested. Can we name the tab "Relative" vs. Defaults. Custom is good.

@vylc, "relative" is not a good name for the first tab because the second one can also be relative, eg, you could have "last Sunday" as the start and "today" as the end time.

@hughhhh hughhhh added the lyft Related to Lyft label May 31, 2018
@betodealmeida
Copy link
Member Author

Video of modifications:

newtimefilter

Screenshots:

screen shot 2018-07-13 at 4 09 06 pm

screen shot 2018-07-13 at 4 09 03 pm

screen shot 2018-07-13 at 4 09 00 pm

@williaster
Copy link
Contributor

looking good, thanks for iterating on/incorporating the feedback!

@mistercrunch
Copy link
Member

LGTM

@mistercrunch mistercrunch merged commit 4fa4163 into apache:master Jul 16, 2018
@mistercrunch mistercrunch deleted the DPTOOLS-324_ui_for_time_filter branch July 16, 2018 21:27
timifasubaa pushed a commit to airbnb/superset-fork that referenced this pull request Jul 25, 2018
* Initial work

* WIP

* WIP

* Working

* WIP

* Still WIP

* Frontend done

* Working version

* Migration working

* Migration working

* Fix freeform rerender

* Remove jquery

* Fix filter

* Unit tests and lint

* Fix py.test

* Improve unit tests

* Ensure freeform is computed at init

* Fix lint

* Trying to fix pyfreeze error

* Remove freezegun

* Address comments

* Use tabs instead of pills

* Regroup options

* WIP

* Change type when clicking calendar

* Fix CSS

* Fix JS lint
wenchma pushed a commit to wenchma/incubator-superset that referenced this pull request Nov 16, 2018
* Initial work

* WIP

* WIP

* Working

* WIP

* Still WIP

* Frontend done

* Working version

* Migration working

* Migration working

* Fix freeform rerender

* Remove jquery

* Fix filter

* Unit tests and lint

* Fix py.test

* Improve unit tests

* Ensure freeform is computed at init

* Fix lint

* Trying to fix pyfreeze error

* Remove freezegun

* Address comments

* Use tabs instead of pills

* Regroup options

* WIP

* Change type when clicking calendar

* Fix CSS

* Fix JS lint
@mistercrunch mistercrunch added 🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels 🚢 0.28.0 labels Feb 27, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
🏷️ bot A label used by `supersetbot` to keep track of which PR where auto-tagged with release labels lyft Related to Lyft 🚢 0.28.0
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

9 participants