Tim Felgentreff wrote:
On 30 Aug 2009, at 05:23, Alexander Lazarević wrote:
failure | FontTest | testFallback
This method apparently a width of the printed character in the failing assertion, however, on tracing the calls I could only find that the tested method, displayString:on:from:to:at:kern: always returns the point passed to it after the 'at:', so this test seems bound to fail this way. I don't know what the meaning of this test is, so I'm not sure what should be done here.
The displayString method should return the advance width. That was broken a while ago in BitBlt (I've pointed this out on vm-dev) but for now I've added a workaround which fixes this issue (and some related, very odd issues when looking at binary stuff in the file list).
failure | FontTest | testResetAfterEmphasized
This test seems to assume the StrikeFont>>reset method empties the array, which it doesn't. Not sure if the test is wrong or the behaviour of the reset method.
The test hasn't been updated. We now also have non-synthetic derived fonts that should not be thrown away upon reset. I fixed that.
failure | MCSnapshotBrowserTest | testNoSelection
Similar with this. The test assumes that when you select nothing in the Snapshot browser an empty string would be displayed, however, it displays the package scripts, or '(package defines no scripts)' if there are none, which is the case in the test. Again, not sure which side to trust here, I changed the test-case and pushed that to the inbox, though.
Thanks for the help!
Cheers, - Andreas