BIND 10 #211: (review) emergency patch: disable /dev/poll in ASIO
BIND 10 Development
do-not-reply at isc.org
Mon May 31 01:24:17 UTC 2010
#211: (review) emergency patch: disable /dev/poll in ASIO
--------------------------+-------------------------------------------------
Reporter: jinmei | Owner: jinmei
Type: defect | Status: closed
Priority: critical | Milestone: 04. 2nd Incremental Release: Early Adopters
Component: Unclassified | Resolution: fixed
Keywords: | Sensitive: 0
--------------------------+-------------------------------------------------
Changes (by jinmei):
* status: assigned => closed
* resolution: => fixed
Comment:
Replying to [comment:1 each]:
> Looks fine, go ahead and merge.
THanks, committed (r1992).
> I must say, I'm starting wonder if using -Werror while also relying on
third-party headers is simply a bad idea... :(
The main problem with ASIO is that we actually include the
"implementation", not just definitions that header files are normally
expected provide. The "implementation" part of ASIO brought these
troubles; otherwise we'd have had the same errors with boost.asio or
sqlite3, etc. So, that's pretty specific to the non boost version ASIO,
not a generic problem of relying on third-party headers with -Werror.
IMO, there shouldn't be an easy excuse to drop stronger warnings or
dropping -Werror (I believe the former is obvious. It should also apply
to -Werror because otherwise I'm sure we'd tend to ignore them).
I belive the middle term solution for this specific problem is to do
essentially the same thing as boost.asio: providing a wrapper layer to
ASIO as a separate internal library and limiting the level of warnings
(or, if there's no other solution, disable -Werror) only within that
wrapper. Other BIND 10 module should still use and benefit from the full
strength of compiler warnings and -Werror.
--
Ticket URL: <http://bind10.isc.org/ticket/211#comment:2>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list