@Öskå I reported this bug to Wolfram and gave every, but really every detail I knew about it. After several mails I got this:
> I have forwarded your question to the developer, but we could not reproduce the same issue. Could you please uninstall Mathematica 10 completely (support.wolfram.com/kb/12473) and reinstall it and see if the issue persists?
@Öskå That's what I write them know too. As long as they don't connect me to a read developer and I always have to write with the support as middle man, I don't think we can track this down.
@Öskå Would you consider writing a bug-report yourself?
So they see that it's not only me having the issue.
:17163465 Hehe.. you shouldn't I pay enough for both of us :-)
@Guillochon You can get a contact form pre-filled with your license number by going to Help -> Give Feedback in MMA. Then you can change the topic. It is worth posting here if it can be answered at all, but if it's something along the line of "what algorithm is used in the code that no end users have access to" then it will probably be closed. You can also try Wolfram Community where there are more WRI employees.
@Szabolcs Hellooo?? But now you can access your FACEBOOK FRIENDS! Isn't that worth something?
:-)
@acl I would be really, I mean really really, surprised if someone at WRI is using their testing framework. Do you really think that with reliable unit tests for all features, something like this wouldn't raise some alerts:
@acl Yep. I find it very interesting that they seem to have changed possible values for Axes without adapting some messages. Again, wouldn't happen with unit tests.
> Doubtless there will be times when the Wolfram System does things you do not expect. But you should realize that the probabilities are such that it is vastly more likely that there is something wrong with your input to the Wolfram System or your understanding of what is happening than with the internal code of the Wolfram System itself.
In other words, "Check your stupid code or stop being a n00b"
> Since the beginning of its development in 1986, the effort spent directly on creating the source code for the Wolfram System is about a thousand developer-years. In addition, a comparable or somewhat larger effort has been spent on testing and verification.
These are merely "vanity numbers" — looks impressive to the newcomer, but doesn't convey any useful information to the experienced. Just like the reputation count on SE...
@rm-rf well in any case, mma is such a ridiculously huge system and it evolves so rapidly that I am not surprised at this. Although some of the bugs are so easy to hit that it makes me wonder.