On Fri, 7 Jan 2022, Michal Prívozník wrote:
On 1/7/22 13:38, Ani Sinha wrote:
>
>
> Ok fine but still, life is not ideal ... libraries do have bugs.
In that case, where do we draw the line? Say pthread has a bug and it
doesn't spawn threads. Worse, it doesn't even return any error value.
Should we mitigate that too?
I'd say stick with what's documented (abort on OOM and/or !NULL
returned) and make our lives simpler. In fact, if we'd crash because we
accessed NULL we will immediately see where and why and can report the
bug to glib for benefit of us and others.
Yes, so if I were you, I would not simply remove the check. I would
replace it with assert().