Update/Modernize Bluetooth Detector Tables

[aharpalaniTO]

Ideally, we should have tables with the following schemas:

• reader_locations: this is a table of all locations at which Bliptrack readers have ever been installed
  • reader_id: unique ID for a unique reader/detector name from the route_points column in all_analyses, this should not be the route_point_id, which is unique for each point of each route.
  • name: e.g. QN_SP (two characters for E/W street, two characters for N/S street)
  • geom
  • int_id: centreline intersection id, for closest intersection. Preferred over px because highway detectors won't be near a signalized intersection, but there could be a logical location that would be an intersection... node
  • date_active: first date at which there was data at this location
  • date_inactive: NULL unless date_last_received is not the most recent date
  • date_last_received: updated daily if the reader is still producing data (if the reader_history associated with that location has been uninstalled)
  • project_name
• reader_history: table of detectors which have been installed at each reader_location
  • reader_history_id: unique ID for the table
  • reader_id: foreign key to reader_locations (location_id)
  • serial_no_bluetooth: device id for bluetooth
  • serial_no_wifi
  • date_installed
  • date_uninstalled
• routes: unique segments on which data is collected
  • analysis_id: analysis id from the all_analyses table
  • name
  • start_street_name
  • start_cross_street
  • start_reader_id: corresponding reader_id from the reader_locations table
  • end_street_name
  • end_cross_street
  • end_reader_id: corresponding reader_id from the reader_locations table
  • geom
  • date_active
  • date_inactive
• bluetooth.reader_status_history : what detectors were active on which date
  • reader_id int, --unique id from reader_locations
  • dt DATE,
  • active BOOLEAN
• ~analyses: some kind of crossover from what's on Blip and what's in our database~
  • ~analysis_id~
  • ~route_id~

There should also be documentation in the Update README on how to update the above tables https://github.com/CityofToronto/bdit_data-sources/blob/master/bluetooth/update/README.md for

[KatiRG]

Created the two tables with the above listed columns.

What rows are supposed to go in them?

[aharpalaniTO]

I'll Slack you the file that will inform what goes into those tables

[KatiRG]

Is the bluetooth_id same as Bluetooth Unit # in the xlsx file? (and similarly wifi_id is WIFI Unit # ?)

[aharpalaniTO]

Yes it is

[KatiRG]

@aharpalaniTO ready for your review. The two tables are under my schema.

Both tables created in python notebooks and pushed to EC2 using an sqlalchemy_mate engine, as documented here: https://github.com/CityofToronto/bdit_team_wiki/wiki/Python#Creating-a-new-table-with-geometry-to-push-to-database

[radumas]

Does this supersede #175 ?

[rickl4]

Can we also tag the PX or int_id to the bluetooth tables?

[aharpalaniTO]

Some weird stuff in the source data - 7 detectors seem to be active with the same Blip ID. I'll need to check with Quin.

[KatiRG]

PX added to cnangini.detectors from gis.traffic_signals by string matching street names.
18 detectors do not have a matching intersection in the gis table:

detector_name	street1	street2
QU_RO	Queen/King	Roncesvalles
A	Gardiner	Wickham
B	Gardiner	Grand
B1	Gardiner	Roncesvalles
C	Gardiner	Dufferin
D	Gardiner	Spadina
E	Gardiner	Parliament
F	DVP	Dundas
G	DVP	Bayview
H	DVP	Don Mills
J	DVP	Lawrence
LB	Harbour EB	Bay
Z2	Lakeshore	Dufferin
Islington	Gardiner	Islington
Spanbridge	DVP	Spanbridge
Beechwood	DVP	Beechwood
Lytton	Yonge	Lytton/Blythwood
Royal York	Gardiner	Royal York

[chmnata]

Found some discrepancy of street names between the detector table and bluetooth.segments. For example, analysis_id 1453197 in bluetooth.segments: street = college and start_crossstreet = parliament, but in cnangini.detectors: street1 is carlton and start_crossstreet = parliament.

These analysis_id in bluetooth.segments do not have the same street name as the detectors table:

analysis_id	street	start_crossstreet	street1	street2	end_street	end_crossstreet	street1	street2	det_start	det_end
1453197	College	Parliament		College	University	College	University		5414
1412913	College	Parliament		College	University	College	University		5414
1453535	Queen	Roncesvalles		Queen	Dufferin	Queen	Dufferin		5412
1454127	King	Roncesvalles		King	Dufferin	King	Dufferin		9002
1454311	King	Broadview		King	Parliament	King	Parliament	5790
1454782	Fort York	York			Bremner	Spadina	Bremner	Spadina		9029
1454747	Fort York	Spadina			Bremner	York	Bremner	York		5417
1427965	Adelaide	Duncan			Adelaide	Sherbourne	Adelaide	Sherbourne	5330
116749	Gardiner	Wickman			Gardiner	Grand	Gardiner	Grand		5815
116858	DVP	Bayview Ramp		DVP	Dundas	DVP	Dundas		5007
116804	DVP	Bayview Ramp		DVP	Don Mills	DVP	Don Mills		5808
1428284	Yonge	Yonge			Lawrence	Jedburgh	Lawrence	Jedburgh		5805
1428295	Yonge	Castlefield		Lawrence	Jedburgh	Lawrence	Jedburgh		5805
1448931	Yonge	Yonge			Yonge	Chatsworth	Yonge	Chatsworth	5615
1461616	Yonge	Castlefield		Yonge	Chatsworth	Yonge	Chatsworth	5615
1448413	Glengrove	Duplex			Yonge	Chatsworth	Yonge	Chatsworth	5615
1448866	Yonge	Lytton			Yonge	Glengrove	Yonge	Glengrove		5619
1448418	Glengrove	Duplex			Yonge	Glengrove	Yonge	Glengrove		5619
1448368	Blythwood	Blythwood			Yonge	Glengrove	Yonge	Glengrove		5619
1448881	Yonge	Castlefield		Yonge	Briar Hill	Yonge	Briar Hill		5613
1448921	Yonge	Lytton			Yonge	Briar Hill	Yonge	Briar Hill		5613
1448363	Blythwood	Blythwood			Yonge	Briar Hill	Yonge	Briar Hill		5613
1460834	Sheppard	Sheppard			Sheppard	Neilson	Sheppard	Neilson		5614
1448557	Morningside	Morningside		Sheppard	Neilson	Sheppard	Neilson		5614
1448745	Sheppard	Sheppard			Sheppard	Murison	Sheppard	Murison		5626
1448552	Morningside	Morningside		Sheppard	Murison	Sheppard	Murison		5626
1448378	Brenyon	Brenyon			Sheppard	Murison	Sheppard	Murison		5626
1448740	Sheppard	Sheppard			Sheppard	Meadowvale	Sheppard	Meadowvale	5621
1448542	Morningside	Morningside		Sheppard	Meadowvale	Sheppard	Meadowvale	5621
1448650	Morningside	Sheppard			Morningside	McLevin	Morningside	McLevin		5625
1448536	Morningside	Morningside		Morningside	McLevin	Morningside	McLevin		5625
1448567	Morningside	Sheppard			Morningside	401	Morningside	401		5611
1448452	Milner	Milner			Morningside	401	Morningside	401		5611
1448531	Morningside	Morningside		Morningside	401	Morningside	401		5611
1448655	Morningside	Sheppard			Milner	Murison	Milner	Murison		5617
1448547	Morningside	Morningside		Milner	Murison	Milner	Murison		5617
1458787	Sheppard	Sheppard			Sheppard	Rouge River	Sheppard	Rouge River	5619
1428129	Sheppard	Herons Hill Way		Victoria Park	Clydesdale	Victoria Park	Clydesdale	5271
1428134	Sheppard	Herons Hill Way		Victoria Park	401	Victoria Park	401		5270
1428103	Sheppard	Herons Hill Way		Sheppard	Pharmacy	Sheppard	Pharmacy		4037
1448799	Sheppard	Sheppard			Sheppard	Pharmacy	Sheppard	Pharmacy		4037
1448760	Sheppard	Herons Hill Way		Sheppard	Settlers	Sheppard	Settlers		5609
1448805	Sheppard	Sheppard			Sheppard	Settlers	Sheppard	Settlers		5609
1448750	Sheppard	Herons Hill Way		Consumers	IBM	Consumers	IBM		5612
1455692	Parliament	Dundas			Parliament	Carlton
1413237	Bathurst	College			Bathurst	Queen
1455061	Bathurst	King			Bathurst	Queen
1454965	Bathurst	Dundas			Bathurst	Queen
1448437	Glengrove	Glengrove			Glengrove	Duplex
1448926	Yonge	Lytton			Glengrove	Duplex
1448887	Yonge	Chatsworth	Yonge	Chatsworth	Glengrove	Duplex			5615
1448432	Glengrove	Duplex			Yonge	Lytton
1448373	Blythwood	Blythwood			Yonge	Lytton
1448856	Yonge	Briar Hill	Yonge	Briar Hill	Yonge	Lytton			5613
1448916	Yonge	Glengrove	Yonge	Glengrove	Yonge	Lytton			5619
1454952	Bathurst	College			Bathurst	Dundas
1455076	Bathurst	Queen			Bathurst	Dundas
1455710	Broadview	Queen			Broadview	Eastern
1455045	Bathurst	Front			Bathurst	King
1454980	Bathurst	Queen			Bathurst	King
1455443	Yonge	King			Yonge	Front
1455724	Broadview	Eastern			Broadview	Queen
1455088	Bathurst	Dundas			Bathurst	College
1413206	Bathurst	Queen			Bathurst	College
1448660	Morningside	Sheppard			Morningside	Morningside
1448457	Milner	Milner			Morningside	Morningside
1448484	Morningside	401	Morningside	401	Morningside	Morningside		5611
1448509	Morningside	McLevin	Morningside	McLevin	Morningside	Morningside		5625
1448675	Sheppard	Meadowvale	Sheppard	Meadowvale	Morningside	Morningside		5621
1448700	Sheppard	Murison	Sheppard	Murison	Morningside	Morningside		5626
1448730	Sheppard	Neilson	Sheppard	Neilson	Morningside	Morningside		5614
1455310	York	Front			York	Bremner
1455538	Jarvis	Front			Jarvis	King
1455510	Jarvis	Queen			Jarvis	King
1455493	Jarvis	Dundas			Jarvis	Queen
1455555	Jarvis	King			Jarvis	Queen
1412908	College	University	College	University	College	Parliament		5414
1453170	College	University	College	University	College	Parliament		5414
1455570	Jarvis	Queen			Jarvis	Dundas
1455213	Spadina	King			Spadina	Queen
1455105	Spadina	Dundas			Spadina	Queen
1455327	York	Bremner			York	Front
1455641	Parliament	Front			Parliament	King
1455616	Parliament	Queen			Parliament	King
1428279	Yonge	Yonge			Lawrence	Mount Pleasant
1428290	Yonge	Castlefield		Lawrence	Mount Pleasant
1448901	Yonge	Chatsworth	Yonge	Chatsworth	Lawrence	Mount Pleasant		5615
1428314	Lawrence	Jedburgh	Lawrence	Jedburgh	Lawrence	Mount Pleasant		5805
1455296	University	King			University	Front
1455231	Spadina	Queen			Spadina	Dundas
116852	DVP	Don Mills	DVP	Don Mills	DVP	Bayview Ramp		5808
116796	DVP	Dundas	DVP	Dundas	DVP	Bayview Ramp		5007
1455195	Spadina	Front			Spadina	King
1448358	Blythwood	Lytton			Blythwood	Mount Pleasant
1455124	Spadina	Queen			Spadina	King
1448871	Yonge	Briar Hill	Yonge	Briar Hill	Blythwood	Mount Pleasant		5613
1448911	Yonge	Glengrove	Yonge	Glengrove	Blythwood	Mount Pleasant		5619
1455676	Parliament	Queen			Parliament	Dundas
1455585	Parliament	College			Parliament	Dundas
1455603	Parliament	Dundas			Parliament	Queen
1455658	Parliament	King			Parliament	Queen
1454293	King	Parliament	King	Parliament	King	Broadview		5790
1453806	Queen	Dufferin	Queen	Dufferin	Queen	Roncesvalles		5412
1454853	Roncesvalles	Queen			Roncesvalles	Dundas
1413812	Dovercourt	College			Dovercourt	Dundas
1454832	Roncesvalles	Dundas			Roncesvalles	Queen
1413817	Dovercourt	Dundas			Dovercourt	College
1428267	Yonge	Yonge			Yonge	Castlefield
1448876	Yonge	Briar Hill	Yonge	Briar Hill	Yonge	Castlefield		5613
1460798	Yonge	Chatsworth	Yonge	Chatsworth	Yonge	Castlefield		5615
1428309	Lawrence	Mt Pleasant	Lawrence	Mt Pleasant	Yonge	Castlefield		5331
1428324	Lawrence	Jedburgh	Lawrence	Jedburgh	Yonge	Castlefield		5805
1454449	King	Dufferin	King	Dufferin	King	Roncesvalles		9002
1455427	Yonge	Queen			Yonge	King
1455456	Yonge	Front			Yonge	King
1455028	Bathurst	Fort York			Bathurst	Front
1454997	Bathurst	King			Bathurst	Front
1454508	Wellington	University	Wellington	University	Wellington	Blue Jays			9031
1455412	Yonge	Dundas			Yonge	Queen
1455468	Yonge	King			Yonge	Queen
1448463	Milner	Milner			Sheppard	Sheppard
1448562	Morningside	Morningside		Sheppard	Sheppard
1448385	Brenyon	Brenyon			Sheppard	Sheppard
1448408	Consumers	IBM	Consumers	IBM	Sheppard	Sheppard			5612
1448794	Sheppard	Settlers	Sheppard	Settlers	Sheppard	Sheppard			5609
1448768	Sheppard	Pharmacy	Sheppard	Pharmacy	Sheppard	Sheppard			4037
1448836	Victoria Park	401	Victoria Park	401	Sheppard	Sheppard			5270
1448815	Victoria Park	Clydesdale	Victoria Park	Clydesdale	Sheppard	Sheppard			5271
1458797	Sheppard	Rouge River	Sheppard	Rouge River	Sheppard	Sheppard			5619
1448494	Morningside	401	Morningside	401	Sheppard	Sheppard			5611
1448526	Morningside	McLevin	Morningside	McLevin	Sheppard	Sheppard			5625
1448685	Sheppard	Meadowvale	Sheppard	Meadowvale	Sheppard	Sheppard			5621
1448710	Sheppard	Murison	Sheppard	Murison	Sheppard	Sheppard			5626
1461700	Sheppard	Neilson	Sheppard	Neilson	Sheppard	Sheppard			5614
1455480	Yonge	Queen			Yonge	Dundas
1455158	Spadina	Front			Spadina	Bremner
1455012	Bathurst	Front			Bathurst	Fort York
1428274	Yonge	Castlefield		Yonge	Yonge
1448906	Yonge	Chatsworth	Yonge	Chatsworth	Yonge	Yonge			5615
1428304	Lawrence	Mt Pleasant	Lawrence	Mt Pleasant	Yonge	Yonge			5331
1428319	Lawrence	Jedburgh	Lawrence	Jedburgh	Yonge	Yonge			5805
1455525	Jarvis	King			Jarvis	Front
1448391	Consumers	IBM	Consumers	IBM	Sheppard	Herons Hill Way		5612
1448783	Sheppard	Settlers	Sheppard	Settlers	Sheppard	Herons Hill Way		5609
1428120	Sheppard	Pharmacy	Sheppard	Pharmacy	Sheppard	Herons Hill Way		4037
1427564	Victoria Park	401	Victoria Park	401	Sheppard	Herons Hill Way		5270
1428097	Victoria Park	Clydesdale	Victoria Park	Clydesdale	Sheppard	Herons Hill Way		5271
1454808	Fort York	Spadina			Bremner	Bathurst
1455243	University	College			University	Dundas
1455385	University	Queen			University	Dundas
1454907	Dufferin	Queen			Dufferin	Dundas
1454937	Strachan	King			Strachan	Queen
1455175	Spadina	Bremner			Spadina	Front
1455142	Spadina	King			Spadina	Front
1455369	University	King			University	Queen
1455256	University	Dundas			University	Queen
1454891	Dufferin	King			Dufferin	Queen
1454867	Dufferin	Dundas			Dufferin	Queen
1454923	Strachan	Queen			Strachan	King
1396053	Lakeshore	Strachan	Lakeshore	Strachan	Lakeshore	Bay			5060
1393437	Lakeshore	Dufferin	Lakeshore	Dufferin	Lakeshore	Bay			5015
1455275	University	Queen			University	King
1455351	University	Front			University	King
116894	Gardiner	Grand	Gardiner	Grand	Gardiner	Wickman			5815
1454879	Dufferin	Queen			Dufferin	King
1455628	Parliament	King			Parliament	Front
1455400	University	Dundas			University	College

with tempa as (select analysis_id, geom, street, start_crossstreet, det1.street1 as det1_st1, det1.street2 as det1_st2, end_street, end_crossstreet, det2.street1 as det2_st1, det2.street2 as det2_st2, det1.bluetooth_id as start_detector, det2.bluetooth_id as end_detector from bluetooth_fixthat2
left join cnangini.detectors det1 on street= det1.street1 and det1.street2 = start_crossstreet
left join cnangini.detectors det2 on end_street = det2.street1 and det2.street2 = end_crossstreet)
select analysis_id, street, start_crossstreet, det1_st1, det1_st2, end_street, end_crossstreet, det2_st1, det2_st2, start_detector, end_detector  from tempa where det1_st1 is null or det2_st2 is null

[chmnata]

PX added to cnangini.detectors from gis.traffic_signals by string matching street names. 18 detectors do not have a matching intersection in the gis table:

detector_name street1 street2 QU_RO Queen/King Roncesvalles A Gardiner Wickham B Gardiner Grand B1 Gardiner Roncesvalles C Gardiner Dufferin D Gardiner Spadina E Gardiner Parliament F DVP Dundas G DVP Bayview H DVP Don Mills J DVP Lawrence LB Harbour EB Bay Z2 Lakeshore Dufferin Islington Gardiner Islington Spanbridge DVP Spanbridge Beechwood DVP Beechwood Lytton Yonge Lytton/Blythwood Royal York Gardiner Royal York

I was looking at joining detector id and analysis_id together for bluetooth highway and I think this is why the text didnt match for these segments: For A: typo for wickham, it should be wickman For G: Bayview is Bayview ramp in segments assuming its 'Bayview Ramp' in gis.traffic.signals too For LB: Harbour EB is Lakeshore in bluetooth.segments I think it would be the same for other unmatched intersections. Is it possible the change the street names in detectors so it would be the same as bluetooth.segments or gis.traffic_signals?

[aharpalaniTO]

@KatiRG - can you check the updated list we just received and identify the following, based on our list of active readers:

a. readers that are currently tagged to the wrong location b. readers that we currently aren't capturing

Thanks!

[KatiRG]

@KatiRG - can you check the updated list we just received and identify the following, based on our list of active readers:

a. readers that are currently tagged to the wrong location b. readers that we currently aren't capturing

Thanks!

So far, I have found that 7 of our detectors have the same BT Unit # but there is no date_removed specified. Of these, 3 can be resolved by matching with the spreadsheet from Jakob.

Will continue to find other discrepancies. More later...

[KatiRG]

Regarding Richmond sensors:

• sensor RM_YO at Richmond and Yonge is currently listed at 43.6519, -79.379295 but should be updated to 43.651478, -79.37895: same intersection but across the street and about 53m away.

• sensor BR2 at Richmond and Sherbourne currently does not have any coordinates in our table, but we now have it from JJ's table: 43.72617, -79.39692. BR2 is marked in our table as having been removed on 2018-08-31 but since it is in JJ's table that means it is active

[KatiRG]

There are two sensors with BT Unit # 5272 in our table:

• 5272 at Castlefield and Yonge: matches location of sensor 5799 in JJ's table within ~35 m
• 5272 at College and Ossington: matches location of sensor 5272 in JJ's table within 10 m

We also have a sensor 5799 but with unknown name, unknown intersection, unknown coordinates, but a known IP address that matches the IP address of sensor 5799 of JJ's table.

Therefore, the Duplicate Status column for the Castlefield and Yonge sensor states the following:
This duplicate should now be Unit # 5799.

[KatiRG]

There are two sensors with BT Unit #5619, located at:

• Intersection = "Sheppard and Rouge River": this one matches JJ's table
• Intersection = "Yonge and Glengrove": this one does not match JJ's table BUT is not listed as having a date_removed. Therefore, the Duplicate Status for this sensor is marked as UNRESOLVED in the final table

[KatiRG]

@aharpalaniTO Sensor DU_BV at Dundas and Broadview (BT Unit # 5620) is listed in JJ's table with coordinates 57.06506, 10.132702, which locates to a building in Vester Hassing, Denmark.

Are we to assume that in this case JJ's table is wrong and that DU_BV is in fact at Dundas and Broadview?

Same is true of sensor 5617 at Milner and Murison, therefore kept original coordinates until otherwise advised!

[KatiRG]

Ready for review by @aharpalaniTO (see email)

[radumas]

Updated the 1st comment based on output from the Blue Tooth Blue Sky meeting

[radumas]

Per discussion with Mohan, think we can skip creating the crossover table between routes and analysis

[radumas]

Per @webgisgeek's investigation, it seems route_point_id pulled from all_analyses aren't actually unique for each detector, so we can't use the values from that table as unique identifiers for our detector-installations (which make up a route).

I've updated the table structure above accordingly.

There should also be documentation in the Update README on how to update the above tables when installing/uninstalling detectors https://github.com/CityofToronto/bdit_data-sources/blob/master/bluetooth/update/README.md for

[webgisgeek]

The start_route_point_id is not unique to a detector. Neither is the end_route_point_id. They are unique for a particular route instead.

analysis_id	report_id	route_id	route_name	start_route_point_id	start_detector	end_route_point_id	end_detector
1455351	1455352	1455330	DT-0115. University-NB_Front-to-King	1455348	KN_UN	1455331	FR_UN
1455296	1455297	1455278	DT-0112. University-SB_King-to-Front	1455279	KN_UN	1455293	FR_UN
1454366	1454367	1454355	DT-0062. King-WB_University-to-Spadina	1454356	KN_UN	1454363	KN_SP
1454196	1454197	1454185	DT-0053. King-EB_Spadina-to-University	1454193	KN_UN	1454186	KN_SP
1454352	1454353	1454343	DT-0061. King-WB_Yonge-to-University	1454349	KN_UN	1454344	KN_YO
1455443	1455444	1455430	DT-0121. Yonge-SB_King-to-Front	1455431	KN_YO	1455440	FR_YO
1454224	1454225	1454213	DT-0055. King-EB_Yonge-to-Jarvis	1454214	KN_YO	1454221	KN_JA
1454340	1454341	1454329	DT-0060. King-WB_Jarvis-to-Yonge	1454337	KN_YO	1454330	KN_JA
1454209	1454210	1454200	DT-0054. King-EB_University-to-Yonge	1454206	KN_YO	1454201	KN_UN
1455427	1455428	1455415	DT-0120. Yonge-SB_Queen-to-King	1455424	KN_YO	1455416	QU_YO
1455468	1455469	1455459	DT-0123. Yonge-NB_King-to-Queen	1455460	KN_YO	1455465	QU_YO

[chmnata]

Suggestions:

• reader_locations: this is a table of all locations at which Bliptrack readers have been installed and are currently deployed
  • location_id: unique ID for a unique reader/detector name from the route_points column in all_analyses, this should not be the route_point_id, which is unique for each point of each route.
  • name: e.g. QN_SP (two characters for E/W street, two characters for N/S street)
  • geom
  • int_id: centreline intersection id, for closest intersection. Preferred over px because highway detectors won't be near a signalized intersection, but there could be a logical location that would be an intersection... node
  • date_active: first date at which there was data at this location
  • date_inactive: NULL unless date_last_received is not the most recent date
  • date_last_received: updated daily if the reader is still producing data (if the reader_history associated with that location has been uninstalled)
  • project_name
  • status: online or offline only
• reader_location_history: table of detectors which have been installed at each reader_location at any given point.
  • reader_history_id: unique ID for the table
  • location_id: foreign key to reader_locations
  • bt_id: device id for bluetooth
  • wifi_id
  • date_installed
  • date_uninstalled
• routes: unique segments on which data is collected
  • analysis_id: analysis id from the all_analyses table
  • name
  • start_street_name
  • start_cross_street
  • start_location_id: corresponding location_id from the reader_locations table
  • end_street_name
  • end_cross_street
  • end_location_id: corresponding location_id from the reader_locations table
  • geom
  • date_active
  • date_inactive
• reader_gaps : gaps of which "we think" detector has no data
  • location_id int, --unique id from reader_locations
  • start_datetime -- start of gap
  • end_datetime -- end of gap

Renaming reader_id to location_id to be more clear, and shrinking the reader_status_history table to reader_gaps table to include dates where it doesn't work instead of the other way around.