On 27/10/25 23:35, Paul Eggert wrote:
> On 2025-10-27 17:33, Collin Funk wrote:
>> Also, I vaguely remember openat2 being discussed on libc-alpha where
>> there was debate over the flags. Did that ever get resolved? Ideally we
>> should use the prototype that glibc adopts.
>
> It hasn't been resolved yet. No solid reason was given on libc-alpha to make
> the new argument read-write so I left it read-only in Gnulib.
I don't have a strong opinion, but since you are pushing for read-only
argument in gnulib having a read-write on glibc only adds extra
complexity.
I was hoping that kernel developers engage more to discuss the need
of a read-write argument, but the idea I got it was added more like
a way to future extensions and a hard requirement.
I will probably send a newer version with a read-only argument.
>
>> I assume you are working for the behavior change for that tar
>> "vulnerability" [1]?
>
> Yeah, people have been bugging me about it.
>
