Extending subsequent tests with dogtail tests?

[Alex ARNAUD]

Le 07/03/2019 à 19:32, Markus Mohrhard a écrit :
 > Most likely no. If we have tests for something it is more likely that it
 > will be fixed after it is discovered by in the end an accessibility
 > regression is similar to any other regression. That means that a test
 > started failing you are expected to inspect the test failure before
 > commiting but we will most likely not stop a release if a new test
 > discovers that something is broken. In general while we do much to avoid
 > regressions and having tests helps we don't treat regressions as
 > complete release blockers. There is always a case-by-case analysis
 > neccessary.

Do you mean someone will break something in LibreOffice and will fix it 
in the future or do you mean someone break accessibility and Hypra has 
to fix it itself in the future ?

In the last scenario, I don't see any reason to have non-regression test 
as we already know that people usually don't take care of accessibility 
and it's why we're forced to keep LibreOffice 4.2 for now. If we expect 
to have one developer fixing bug of thousand of others (on many free 
software) especially because it's accessibility I don't think it's reliable.

Best regards,
Alex.