This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCH] Fix ether_ntoa() output format
- From: Roland McGrath <roland at hack dot frob dot com>
- To: OndÅej BÃlka <neleai at seznam dot cz>
- Cc: mokhtar <mbmessaoud at gmail dot com>, libc-alpha at sourceware dot org
- Date: Wed, 23 Oct 2013 16:05:16 -0700 (PDT)
- Subject: Re: [PATCH] Fix ether_ntoa() output format
- Authentication-results: sourceware.org; auth=none
- References: <op dot vggmcoovgweiyn at localhost dot localdomain> <4C5190C2 dot 7020801 at redhat dot com> <1381613691126-247250 dot post at n7 dot nabble dot com> <20131013091354 dot GA8644 at domone dot podge>
This change may need further justification. It is the de facto norm that
MAC address notation is accepted with one-digit fields, so there is no
particular reason to think that the existing behavior causes any problems.
Conservatism/paranoia says don't perturb in fear of unintended effects on
existing users. The middle road would be to change with a new symbol
version and keep the existing behavior for existing binaries.
OTOH, recent FreeBSD uses %02x format so it's likely that the change won't
cause trouble. So I don't really have an objection.
The log entry is inadequate. Pretty much any entry that says "fix foobar"
is wrong. It should describe the change, which is very simple to describe
in this case ("Use %02x rather than %x for each address byte.").
This user-visible change needs a BZ report (which should give
justification, preferably something practical rather than just a vague
reference to a non-binding not-really-a-spec) and BZ# tag in its log entry.