[TUHS] Version 256 of systemd boasts '42% less Unix philosophy' The Register

[Warner Losh]

On Fri, Jun 21, 2024 at 10:40 AM Henry Bent <henry.r.bent at gmail.com> wrote:

> Sure, and I don't disagree.  I was just using an old OS to make a point
> about corner cases; it would be just as applicable if I had a modern OS
> that for whatever reason lacked strdup(), or your personal favorite "but
> everyone has this!" function.  You're not going to be able to cover all
> bases all the time, and I'm sure that there are plenty of code authors who
> aren't interested in formally supporting anything outside of the most
> common operating systems.  If their autotools-based projects work on my
> other OS that's great, but it isn't the fault of autotools if the project
> isn't coded with my OS in mind.
>

Normally in modern software, "has it or not" is controlled by some
pre-processor variable you can check. The problem comes in when you have
under-conformant systems that claim conformance with POSiX 1-20xx, but that
lack that one interface mandated by it (and one that's not controlled by
some other thing... posix is super complex, for good and for ill). And you
also have the edge case of "newly defined in C11" say, and the base
compiler doesn't claim C11 conformance, but this function is none-the-less
available. It's really really hard to know if it's there or not w/o testing
for it. That even goes for "it's a linux box" since musl vs glibc has
variations that you won't know about until you check.

So it doesn't have to be something that should be as ubiquitous as strdup
to run into issues.

Warner
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.tuhs.org/pipermail/tuhs/attachments/20240621/82d05226/attachment.htm>