[mkgmap-dev] Test failure in r2448

Tue Jan 15 08:05:57 GMT 2013

Hi Steve,

Steve Ratcliffe wrote
> I've re-enabled the nightly builds but the last one is 2447 as from 2448
> there are test failures.
> 
> One is because the roads are now added right at the end and the
> test does not call the converted end() method. The attached
> patch fixes that.

Thanks. I should have found this myself, but the tests produce so many
failure messages that I did not notice this additional one.
Or maybe it is not normal to have 16 failures?
This is my alltests-fails.html:
alltests-fails.zip
<http://gis.19327.n5.nabble.com/file/n5744571/alltests-fails.zip>  

Steve Ratcliffe wrote
> The other is partially the same, but it appears that the resulting
> ways are also removed by the removeShortArcs routine.

Hmm, that one is difficult because the patched test does something
completely
different compared to the ElementSaver:
ElementSaver calls converter.convertWay() for all ways and then finally
converter.end()
The patched StyleTester.runSimpleTest()  calls xyz.convertWay() directly
followed by xyz.end() for each way.

The failing test could be changed to use non-road lines, but I think this
not what we want.

The javadoc in interface OSMConverter says that convertWay() should do the
conversion,
but with r2448 a part of the conversion is done in end() , so the whole
program flow 
was changed.

Should I try to change the tests so that they work like ElementSaver ?
That would mean we can only test the result after calling the final end()

Or do you see a better place for the removeShortArcs() routine?
I thought about putting it into MapBuilder, but at this place we would not
be able to report warning with the Way.toBrowseURL() method.

Ciao,
Gerd

--
View this message in context: http://gis.19327.n5.nabble.com/Test-failure-in-r2448-tp5744543p5744571.html
Sent from the Mkgmap Development mailing list archive at Nabble.com.