Christoph Buchner

> we should update the display unit accordingly. not the displayunit in the source code, though, right?

> > pyfmi install via pip package is too old it seems it has pyfmi version 2.5 which seems to be very old. > > I think we already noticed...

> The question is: do we prefer to get some "false positive" warnings, such as dimensional inconsistency warnings from the above-mentioned examples, but also to catch plain wrong equations such...

I can't answer the resource question, but would that really be an argument to _not_ implement better checking :interrobang: Because doing it would point out too many errors for "free"?...

Thanks for that zip file, that is useful! I count/grepped 338 instances of incompatible units. It might be that there are some real false positives in that log, though. For...

> As I understand, this is not a false positive, but a tool issue. Sorry if this was not clear enough. This is what I meant, a false detection by...

> Well, fractional powers aren't really specified - and you cannot specify them for variables. Thanks, interesting, I was not consciously aware. Still, physically/mathematically it's ok here (simpler example: `sqrt(vel1)*sqrt(vel2)`...

> Also considering addition and subtraction seemed to only give false positives(*), and relational expressions just gave lots and lots of false positives. Requiring users to change `r+1e-12` or `abs(r)>1e-12`...

> The updated idea is that we shouldn't consider all numbers _appearing_ in expressions with other operands. > > Instead consider specific cases and say that it has unit="1" when...

I've documented scaling/dpi symptoms which are probably related to these over in #12523 (in my case triggered by a multi-monitor setup, not ultrahigh resolution).