On Tuesday, August 21, 2012 12:37:10 PM UTC+1, Ben Hood wrote: > > > So I'm wondering what practical options are left - I was thinking >> something >> > along of the lines of detecting the dialect at runtime and issuing a >> HSQLDB >> > specific syntax for the purpose of the test suite. Not ideal, but it >> seems >> > to be the least worse solution. > > Interestingly enough, this minor tweak to the query construction _appears_ to have solved the HAVING issue with HSQLDB: https://github.com/lshift/diffa/commit/c5e69ece134232a7b6b616d187e3ecdc15f564af
I'm going to investigate this a bit further to see whether this works on Oracle and MySQL. Cheers, Ben
