[Libreoffice-qa] minutes of ESC call ...
bjoern.michaelsen at canonical.com
Thu Oct 9 10:30:00 PDT 2014
On Thu, Oct 09, 2014 at 04:35:12PM +0100, Caolán McNamara wrote:
> Wouldn't it be nice if the bugs could be split into two categories of
> 100% reproducible and not, and script the 100% reproducers and
> automatically re-run the script on each milestone to see which ones got
> fixed in passing. Though I guess in practice that would depend on
> reliable gui automation in a lot of cases.
Hmm, I kinda like that. Just wondering: Maybe we should have an EasyHack to
write tests for these regressions? At least some of those should be possible to
recreate in good test code. For one, this would give us relevant information:
1/ Is this regression really still around on master
2/ Should allow full-auto bibisect to the commit introducing the trouble
3/ Is this a Heisenbug? Does it happen on all platform? etc.
And fianlly: We will have a test in the end even after this is fixed. After
all, we all know that the motivation to write a test drops even harder once
there is a fix as "this is not a problem anymore".
More information about the LibreOffice