Nikita Kniazev

Thanks for the catch, I pushed the fix into the branch.

> I have another example of some unexpected behaviour. > > This is based on code you supplied when exploring #679 > > ... > > I am still not...

> But the code I supplied **does** compile on a machine with these versions: It only confirms that it is not related to the changes we are discussing here. >...

File a new issue for it (not a regression and not related to this one).

I heard of the issue and was actually looking into a reproducer. > In general, you need an accumulator that can provide more precision than a type T provides Not...

> > I heard of the issue and was actually looking into a reproducer. > > > In general, you need an accumulator that can provide more precision than a...

> Anyway, the fmt library only ouputs the number as -0.0009308399603385636. I wonder if that actually would be correct according to the IEEE 754 and the C++ standard. fmt gets...

> IIRC, the accumulator has the same type (i.e. double in this case). No, Qi uses integer accumulator since https://github.com/boostorg/spirit/commit/6832dc229872d7f9573a506cfe1ed640dc79871e > @Kojoley, if you have time to track this issue...

> > > @Kojoley, if you have time to track this issue down with a repro, that would be great. > > > > > > I already tracked it...

> The problem appears when the grammar is wrongly designed which causes it to be defective by design. Those grammars are not defective, but just left recursive (as further noted...