BIND 10 #901: Alter logging interface to use positional formatting for message arguments
BIND 10 Development
do-not-reply at isc.org
Fri May 6 14:02:28 UTC 2011
#901: Alter logging interface to use positional formatting for message arguments
-------------------------------------+-------------------------------------
Reporter: | Owner: vorner
stephen | Status: reviewing
Type: | Milestone:
enhancement | Sprint-20110517
Priority: major | Resolution:
Component: | Sensitive: 0
logging | Sub-Project: DNS
Keywords: | Estimated Difficulty: 0.0
Defect Severity: N/A | Total Hours: 0
Feature Depending on Ticket: |
Add Hours to Ticket: 0 |
Internal?: 0 |
-------------------------------------+-------------------------------------
Changes (by stephen):
* owner: stephen => vorner
Comment:
> Well, there's a need when it's returned from the error() and such
function (return from C++ function needs a copy constructor) and we can't
really use a reference there.
True, I hadn't considered that. And of course you have to cope with the
destruction of the original object... There are some subtleties here, so
it might be worth adding a paragraph to the header to explain that.
> And I wasn't sure if it's OK regarding the lifespan of temporary
objects...
I think it would be OK - a discussion of this issue can be found
[http://stackoverflow.com/questions/584824/guaranteed-lifetime-of-
temporary-in-c here].
''(regarding missing parameters...)''[[BR]]
Leave it for now, but I suggest that we take this to the bind10-dev list.
> (regarding semi-colon in empty braces)
Again leave it. If we get an error during a compilation, we can add it
later.
''(regarding including macros.h in logger.h)''[[BR]]
Including logger.h in macros.h would be OK. I think that include files
should be "stand-alone" - if A requires that B be included prior to it,
why doesn't A include B?
> Well, if there's no output, then the other tests would crash, so we
can't continue. But if there are more outputs, we want to continue and
check the value of the output, but still report there are too many. Maybe
this is little bit overcomplicated. Should I just replace it with
ASSERT_EQ and not care about the output in such situation?
I think so. If it is producing more than one output then there is some
problem that needs fixing. If subsequent tests also showed up a problem,
it is quite possible that they are related to the first one.
!ChangeLog entry is OK.
--
Ticket URL: <http://bind10.isc.org/ticket/901#comment:6>
BIND 10 Development <http://bind10.isc.org>
BIND 10 Development
More information about the bind10-tickets
mailing list