BIND 10 #61: review: dnstime
BIND 10 Development
do-not-reply at isc.org
Sat Feb 19 01:44:21 UTC 2011
#61: review: dnstime
-------------------------------------+-------------------------------------
Reporter: jinmei | Owner: jinmei
Type: task | Status: accepted
Priority: major | Milestone: A-Team-
Component: | Sprint-20110223
DNSPacket API | Resolution:
Keywords: | Sensitive: 0
Estimated Number of Hours: 2.0 | Add Hours to Ticket:
Billable?: 0 | Total Hours:
Internal?: 0 |
-------------------------------------+-------------------------------------
Comment (by jinmei):
Branch trac61 is ready for review.
The code change itself should be pretty trivial. Test cases may be
subtle, but I tried to give detailed explanation about what's tested.
I also gave a detailed description to the functions (which was simply
missing so far).
In the code, one possibly controversial point would be that I replaced
the use of gmtime() to in-house conversion logic derived from BIND 9.
One reason for this is because (as commented in the code) gmtime()
may not work for 64bit integers (depending on systems). Another
undocumented reason is gmtime() is not thread safe, and b10-resolver
may want to have thread safety here (there's gmtime_r(), but I've seen
cases where _r() functions are not available).
This is the proposed changelog entry:
{{{
173.? [bug]* jinmei
src/lib/dns: revised dnssectime functions so that they don't rely
on the time_t type (whose size varies on different systems, which
can lead to subtle bugs like some form of "year 2038 problem").
Also handled 32-bit wrap around issues more explicitly, with more
detailed tests. The function API has been changed, but the effect
should be minimal because these functions are mostly private.
(Trac #61, git TBD)
}}}
--
Ticket URL: <http://bind10.isc.org/ticket/61#comment:13>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list