On 09/15/2010 02:17 PM, Justin Clift wrote:
On 09/16/2010 05:49 AM, Eric Blake wrote:
<snip>
+MacOS 10.5.
Is "10.5" the right choice for this?
Wondering if maybe it should be a range indicator, as "10.6"(.4)" here
has this issue, and it sounds like pretty much all of the previous
versions will too.
Our general policy with
gnulib/doc/{glibc,posix}-{headers,functions}/*.texi has been to document
the most recent version that has been explicitly tested to have the
issue. It's couched in open-ended terms (the failure exists on at least
this modern porting target) rather than closed-ended (only these targets
have the issues), and it is generally a safe bet that older versions
likely have the same bugs. Whether newer versions have fixed the bug is
irrelevant in portability matters - really, once you fix an issue for
any one sane porting target, you've avoided the problem for all other
porting targets even if you have not yet verified whether those other
porting targets share the same issue.
So I think the wording is okay as-is. But thanks for the scrutiny :)
Meanwhile, over time, some porting targets are no longer sane (for
example, this month there has been a question of whether IRIX 5.3
support still makes sense, now that IRIX 6.5 is top-of-the-line and the
vendor no longer maintains patches for the older version). Eventually,
when targets are retired, we could sweep back through the collected
wisdom of portability notes and cull out old platforms; at that point
we'd have to check if the problem still exists on current platforms
before pruning corresponding code from the original workaround.
--
Eric Blake [email protected] +1-801-349-2682
Libvirt virtualization library http://libvirt.org
