Those tests had been disabled due to running too long, given the high
number of permutations created.
A possible solution would be to run the test only on a random subset of
those. 5% on my machine results in 2 seconds of run, which might be more
acceptable.
Restored version additionally tests both C locale and system locale.
Details
Details
- Reviewers
nienhueser rahn - Group Reviewers
Marble - Commits
- R34:82fab3746713: Restore TestGeoDataCoordinates' testFromStringDMS & testFromStringDM
All tests pass
Diff Detail
Diff Detail
- Repository
- R34 Marble
- Branch
- restoreDMSAndDMParseTests
- Lint
No Linters Available - Unit
No Unit Test Coverage
Comment Actions
With LonLatParser ported from QRegExp to QRegularExpression, the full set of DMS/DM tests (so not only random 5% subset) here now take ~ 25 secs instead of ~49 secs on my Intel(R) Core(TM) i5-2520M CPU @ 2.50GHz.
Still quite some time, but perhaps no longer too painful?
Comment Actions
There might have been some discussion off-page about this patch, but I do not remember any details. Would you like this to go in, or shall I discard it?
In terms of reproducability given the random skipping, if a test run has a falling test, it also shows the data line on which it fails, so can be explicitely invoked when one goes to fix things.