29 dec. 2020 kl. 21.01 skrev Eli Zaretskii : > IMO, making this kind of changes is just asking for trouble: the gains > are null and void (I don't see why we should care about consistency > here), while the potential for breaking something out there is very > real. The gains are not null and void (or the change would not have been proposed). There is no evidence for it being likely to break anything, but we can make the reform optional and turned off by default. That way, code for which it is desirable and safe could bind a dynamic variable during the pretty-printing. Updated patch attached.