On 2016-09-08, Scott Kostyshak wrote: > On Thu, Sep 08, 2016 at 07:02:52AM +0000, Guenter Milde wrote:
>> These tests are inverted, because compilation fails here. > I see. >> If the inverted tests "fail to fail" at your site, the lines ... >> should be moved from suspiciousTests to unreliableTests under >> Sublabel: varying_versions > Ah I forgot about that. And I suppose the inversion status should depend > on the newest version. For example, if there are different test results > based on different statuses of TeX Live packages, then the test should > pass for the user with the updated TeX Live 2016, right? While I agree with this in theory, actually the labeling mechanism is constructed so that inversion takes precedence and prevents tagging with "unreliable". Maybe this can be changed in an overhoul of the test setup... Günter
