upgrade OSHDB to version 0.6

[tyrasd]

see https://github.com/GIScience/oshdb/blob/master/CHANGELOG.md#060

• "fix" tests after upstream changes:

  improve accuracy of built-in geometry helper functions which calculate the geodesic lengths and areas of OSM geometries

• double-check new integrated filter implementation
• add to changelog

[joker234]

@tyrasd you could update the ohsome parent as well ;)

[tyrasd]

@tyrasd you could update the ohsome parent as well ;)

good point, done in c5b9f56

[tyrasd]

@bonaparten @FabiKo117 I quickly checked the failing assertations from https://jenkins.ohsome.org/job/ohsome-api/view/change-requests/job/PR-86/3/consoleFull and found all "new" results to be within a 0.35% (relative) margin of the "old" test-values. Which is what I would have expected from the changes in GIScience/oshdb#193. Did I overlook some cases where the discrepancy is larger?

[FabiKo117]

@bonaparten @FabiKo117 I quickly checked the failing assertations from https://jenkins.ohsome.org/job/ohsome-api/view/change-requests/job/PR-86/3/consoleFull and found all "new" results to be within a 0.35% (relative) margin of the "old" test-values. Which is what I would have expected from the changes in GIScience/oshdb#193. Did I overlook some cases where the discrepancy is larger?

@tyrasd No, it seems like this is the maximum discrepancy then. Probably I was thinking of the higher differences in bigger values, when I was referring to differenes in the lower percentage area when mentioning it in the meeting. I'd suggest to use a delta of 0.5 in the tests + adapting the current values to the more accurate ones. What do you think about that?

Another thing: all the /contribution tests are now failing for the following reason:
groupByEntity() must be called before any `map` or `flatMap` transformation functions have been set
The first line in the ohsome API code that is referenced in the error message is this one.

[tyrasd]

I'd suggest to use a delta of 0.5 in the tests + adapting the current values to the more accurate ones. What do you think about that?

that sound good 👍

Another thing: all the /contribution tests are now failing

oh, true. let me debug that.

[tyrasd]

Another thing: all the /contribution tests are now failing

oh, true. let me debug that.

Quick update: I run into the following bug while trying to fix this: GIScience/oshdb#321 😬

[tyrasd]

Another thing: all the /contribution tests are now failing

oh, true. let me debug that.

Quick update: I run into the following bug while trying to fix this: GIScience/oshdb#321

this should now hopefully be fixed with oshdb version 0.6.1 (updated this PR with 6e66a5e) 🤞

[bonaparten]

Tests are now updated with new values and delta.

[FabiKo117]

Please look at the detailled review comments. There are several unused import statements from removed code parts, make sure to check again the changed classes and organize the imports there again (think mostly in classed edited by @tyrasd).

About the tests: I did not check every single test, but individual ones without the delta and they worked fine and all tests are running through of course. Please check the review comment about commenting @bonaparten.

[FabiKo117]

See inline comments.

[FabiKo117]

Looks good now. Marking the test values as 'final' also makes sense here.

[joker234]

Looks good now. Marking the test values as 'final' also makes sense here.

I checked 3d3ec09 against a test cluster with oshdb version 0.6.1 using our usual tools and couldn't find anything suspicious.