>>>>> On Thu, 21 Sep 2023 09:33:57 -0700, Paul Eggert said:
>
> Could you please look at the current sources? That part of 'tar' has
> changed significantly since 1.35, and I don't see a bug in the latest
> version. If the bug's still there, please let us know.
The latest version looks correct to me too, but I think it would crash
if the final node in the chain has the _WSNF_JOIN flag. If this is
supposed to be impossible, then the tests for non-NULL p at lines 571
and 575 are redundant and confusing.
571 for (p = node; p && (p->flags & _WSNF_JOIN); p = p->next)
572 {
573 len += wsnode_len (p);
574 }
575 if (p)
576 len += wsnode_len (p);
577 end = p;
__Martin
