Luigi
Sorry for the delay. I have to postpone it, no time now . I am not sure that I will fix it because I don't see a bug (yet).
clearly it is a borderline case and I know that Hans et al do not see log compatibility of any importance and from a certain viewpoint I can understand it. However, for folks that attempt to ensure that TeX code works correctly across different engines it makes life quite hard, because any change in the log output means you can't certify some outcome as correct and then make sure that you get the same output on the next engine. For LaTeX we already have a larger script that tries to mask most differences, but some are difficult to if not impossible to catch late in the game (especially if they involve spaces) which means we then end up with separately certified logs and that s a nightmare to maintain (and leads to overlooking changes that are relevant). The other type of issue with that (and which is the starting point of this thread) are GUI interfaces that try to interpret errors and warnings. Of course they can maintain different code branches based on engine used, but again that leads to complications which are better solved upstream but not producing differences when not necessary. So no, they aren't necessarily bugs if the log output differs but it would be nice and quite helpful if they don't (unnecessarily) when possible. cheers frank