I agree with Vladimir, and am inclined to stop the vote and make a new RC with a fix to this bug. As a test issue, it is not “serious”, but it does affect people’s perception of Calcite. And since the fix only affects tests, we will not need to re-do compatibility testing.
Vladimir, Can you provide a pull-request? Or if you prefer, you can commit to branch-1.6. Julian > On Jan 18, 2016, at 12:43 AM, Vladimir Sitnikov <[email protected]> > wrote: > > Adding a dummy `DriverManager.getDrivers();` to > AlternatingRemoteMetaTest.<clinit> resolves the issue. > > static { > try { > DriverManager.getDrivers(); // <-- added a line to initialize > DriverManager class > DriverManager.registerDriver(new AlternatingDriver()); > } catch (SQLException e) { > throw new RuntimeException(e); > } > } > > So the question is if we are fine to ship without such a modification. > Initially I thought it was not a big issue, however, it turned out to > reproduce consistently for me, thus I expect it might impact lots of > users who build Calcite. > > Can we sneak DriverManager initialization in 1.6.0? > > After the modification, it works fine for me. Tests pass, > mat-calcite-plugin works. > > Vladimir
