Pavel Sanda <sa...@lyx.org> writes: >> I think this test is wrong. I am not sure how to make such a thing >> work... > > hmm i thought this test means that there was a day when > convert<string>(1.1) actually returned 1.1 and this test just shows > there was some change underneath... > > i'm not sure whether we handle for example custom lengths as strings > or numbers, but in the second case some problems might arise?
Testing the exact representation of a real number is asking for this kind of instability. If we were testing a conversion with a given (not too large) precision, that would be OK. I am not sure what to do, besides dropping the test. JMarc
