Index | Thread | Search

From:
"Theo de Raadt" <deraadt@openbsd.org>
Subject:
Re: Documentation in ar(5) does not reflect ar(1) behavior
To:
Ingo Schwarze <schwarze@usta.de>
Cc:
Sören Tempel <soeren@soeren-tempel.net>, tech@openbsd.org
Date:
Mon, 05 Oct 2026 09:39:38 -0600

Download raw body.

Thread
Step back and look at how this happened.

In practice, ar had two uses.

One use was file-tree transport, and that use was replaced in the sysv world
with cpio and in the bsd workd with tar, and eventually with tar variants
(mostly because the cpio commandline options sucked ass more than the tar
commandline optins.

The other use was static library .a files.  The ar file format was OK
for this for a while, but bursting at the seams.

Since the first use case had fallen out of use, I suspect the toolchain
people incrementally started making mostly undocumented changes and
extensions to the ar format?

Everything follows from that.  I do not think the manual page needs
to tell these stories.  manual pages should avoid telling lies, and
if they can't tell an accurate truth and therefore they sound hollow
that's also fine.