bug#55395: What does (1 2 3 . #2) mean?

["Mattias Engdegård" <mattiase@acm.org>]

18 maj 2022 kl. 23.16 skrev Stefan Monnier <monnier@iro.umontreal.ca>:
> 
>> [(1 2 3 . #0)]
>> 
>> could mean either
>> 
>> [#1=(1 2 3 . #1#)]
>> 
>> or
>> 
>> #1=[(1 2 3 . #1#)]
> 
> AFAIK it *should* mean the latter (#0 is the root of the overall
> printed object, not just the list).

Yes, but the code was already a bit ambiguous in that respect. We don't keep track of the cons-level depth when printing; print_depth treats each list nesting as one step regardless of where in the list the nesting occurs. I suppose this could be remedied but I'm not going to work on it right now.

The previously attached patch has been pushed to master, on the grounds that it should be strictly better than what we had before; correctness should be back at the level before the change in circularity detection algorithm broke the tail index completely.

> Yup.  But we don't shy away from playing the fools.

You know me, I don't even need to play one. It all comes natural.