Re: Regression tests fail with musl libc because libpq.so can't be loaded

On Wed, Mar 20, 2024 at 02:12:54PM +1300, Thomas Munro wrote:
> On Wed, Mar 20, 2024 at 12:54 PM Bruce Momjian <bruce(at)momjian(dot)us> wrote:
> > So this would truncate the process title on all Linux that have an LD_
> > environment entry, even those without musl?
>
> Yep. How long is /proc/XXX/cmdline (check with wc -c /proc/...) in a
> postmaster near you? You'd always get that much, plus as much of

$ cat /proc/2000/cmdline |wc -c
30

> /proc/XXX/environ as we can find before you reach LD_XXX=, which on a
> typical system would, I guess, usually be never. If it's a problem
> you could try to arrange for LD_ XXX to come later in environ[]. What
> I observe is that they seem to get copied in backwards, wrt the
> environment exported by the parent, so if you set DUMMY=XXXXXXXX just
> before starting the process it'll make sacrificial space in the right
> place (but I'm not sure where that effect is coming from so don't
> quote me).

I am just cautious about changing behavior on our most common platform
for a libc library I have never heard of.

--
Bruce Momjian <bruce(at)momjian(dot)us> https://momjian.us
EDB https://enterprisedb.com

Only you can decide what is important to you.