On Fri, Mar 23, 2012 at 07:26, Robert Newson <[email protected]> wrote: > I'm revising that, benoit rightly suggests adding a create_if_missing > option which would work the way randall thought create worked.
With Benoit and Bob's help, we decided that log at debug was fine and I added a log for all uncaught exceptions that would return 500s to the client to be sure things don't go unnoticed by the admin in production. Should be unblocked now for 1.2 release. -R > > B. > > On 23 March 2012 14:08, Robert Newson <[email protected]> wrote: >> http://git-wip-us.apache.org/repos/asf?p=couchdb.git;a=commitdiff;h=1d13adcdcc9e31b80c8c2c4d84bddbe8452e20ec >> >> On 23 March 2012 14:04, Noah Slater <[email protected]> wrote: >>> \o/ >>> >>> On Fri, Mar 23, 2012 at 1:56 PM, Robert Newson <[email protected]> wrote: >>> >>>> Working on it. >>>> >>>> On 23 March 2012 13:54, Noah Slater <[email protected]> wrote: >>>> > I am available tonight to try and ship 1.2.0. Will we be ready by then? >>>> > >>>> > On Fri, Mar 23, 2012 at 1:39 PM, Robert Newson <[email protected]> >>>> wrote: >>>> > >>>> >> Randall, >>>> >> >>>> >> I could write a whole thing here but I'll cut it short and simply >>>> >> apologise for any offense you have taken, it was not intended. My >>>> >> 'chastisement' was intended humorously, but you were essentially >>>> >> saying that your fix should have worked if only [create] worked the >>>> >> way you imagined it did rather than how it does (which implies you >>>> >> didn't try it). >>>> >> >>>> >> I'll state again that logging at debug level is not a great solution. >>>> >> We should log at error level *except* when auto-creating the _users >>>> >> and _replicator dbs. It's a more involved fix, but I think it's worth >>>> >> it. >>>> >> >>>> >> B. >>>> >> >>>> >> >>>> >> On 23 March 2012 13:28, Randall Leeds <[email protected]> wrote: >>>> >> > On Fri, Mar 23, 2012 at 06:15, Robert Newson <[email protected]> >>>> wrote: >>>> >> >> I'm also -1 on your revised solution. We go to the trouble of >>>> >> >> carefully logging and formatting these errors and then log them at a >>>> >> >> level that approximately no one ever runs at (debug is far too noisy >>>> >> >> to use in production, for instance). >>>> >> > >>>> >> > s/we/I/, and that's the point of debug. >>>> >> > >>>> >> >> >>>> >> >> B. >>>> >> >> >>>> >> >> On 23 March 2012 13:14, Robert Newson <[email protected]> wrote: >>>> >> >>> " Is there a good reason why we don't honor >>>> >> >>> the create option in the way I expected?" >>>> >> >>> >>>> >> >>> Is there a good reason you committed a fix to a release branch >>>> without >>>> >> >>> testing it? >>>> >> > >>>> >> > Are you referring to the different fix which I quickly reverted on >>>> >> > 1.1.x? I did test that I didn't get your spurious error report, but >>>> >> > tested more before committing to 1.2.x, caught my oversight, reverted, >>>> >> > and offered the only solution which is non-invasive. >>>> >> > >>>> >> > Is there a good reason you'd prefer to chastise me than to answer my >>>> >> > question? I woke up at 5:30am and checked my e-mail because my >>>> >> > roommate happened to be in there already. I decided I felt guilty >>>> >> > enough about a spurious log message and cared enough about shipping >>>> >> > 1.2 to stay awake and investigate. I committed a fix because I was >>>> >> > trying to be _helpful_. >>>> >> > >>>> >> > I'm asleep for a few hours. Happy to discuss process when I return. >>>> >> >>>>
