This is the mail archive of the
mailing list for the glibc project.
Re: RCS and SCCS keywords
- From: Roland McGrath <roland at hack dot frob dot com>
- To: Florian Weimer <fweimer at redhat dot com>
- Cc: GNU C Library <libc-alpha at sourceware dot org>
- Date: Wed, 27 Apr 2016 13:37:48 -0700 (PDT)
- Subject: Re: RCS and SCCS keywords
- Authentication-results: sourceware.org; auth=none
- References: <57211C65 dot 1020401 at redhat dot com>
> Even if we remove them now, they are still available from the history.
> Their immediate benefit is unclear because I expect that many of these
> files have been edited before they were imported into glibc, so these
> markers do not indicate an unambiguous baseline of a file anyway.
In all cases I can think of, such files were imported verbatim and so
the markers were in fact clear indicators of their original source
(inasmuch as those markers were kept up to date in the 4.4BSD and ISC
source trees, which I think they were). For many years, we were still
taking wholesale updates to things like the resolver code from BSD/ISC
sources. But all of our code diverged many years ago and it has been
a long time since we considered it even close to possible (or
sensible) to merge back in changes from the "originals". So I don't
think they give us any value at all. I'm much less sure about the
libm (mostly in sysdeps/i386/fpu and sysdeps/ieee754) cases, but I
strongly suspect we have permanent divergence there too.
I'm inclined to suggest a single massive removal of all of these,
rather than just getting them out piecemeal as you clean up specific
files in other ways. At least, you might as well do all of resolv/
in one swell foop.