On 09.02.2014 02:34, Ben Reser wrote: > On 2/8/14, 2:42 PM, Branko Čibej wrote: >> Many tests fail on OSX with SQLite 3.8.3. See the attached log. > I ran a set of tests with OS X 10.9.1 and SQLite 3.8.3 amalgamation and it > completed successfully. I notice a lot of your tests have an error or warning > about write-locks. Is this possibly something wrong with your test > environment?
I didn't use the amalgamation, but the pre-built version from Homebrew, which does work for me on trunk. Also, the pre-built 3.8.2 worked with 1.8.5. It could be a problem with the Homebrew sqlite version, or it might not. I'm still trying to track this down, but until I do, I though it would be safer to report the possible blocker than to go ahead with a broken release. >> - 'make swig-py and 'make swig-rb' do not work in out-of-tree builds >> with the generated wrappers included in the tarball. >> (N.B.: In 1.8.5, the included Swig wrappers did not work at all.) > Oops, I fixed this with r1543187, but apparently it got forgotten to be > nominated so didn't make it in 1.8.6. > > I'm assuming N.B. means non-blocking here, so if we can figure out the problem > with SQLite 3.8.3 you'd be +1 then? It's "nota bene", not "non blocking" ;) but this is indeed an improvement, and the SQLite 3.8.3 problem is the only reason for stopping the release for now. -- Branko Čibej | Director of Subversion WANdisco // Non-Stop Data e. [email protected]
