Hey,
On Fri, Mar 8, 2019 at 6:33 PM Miklos Vajna <vmiklos@xxxxxxxxxxxxx> wrote:
Hi,
The current practice is: if 'make check' passes (which is more or less
enforced by Jenkins) and the change looks good to a reviewer, the change
goes in. And then releases are based on time, so it's really rare that
there are "blocker" bugs which would delay a release.
The ideal for any new kind of testing (including accessibility) is that
it's integrated into 'make check', and whatever those tests cover are
not OK to be broken anytime.
If the proposed a11y tests are part of make check, then it's easy to
promise that they won't be broken; otherwise it's just a best effort
thing without any guarantees.
At least that's how I understood what Markus wrote, and I agree with
that.
Exactly that. With the added comment about the reliability necessary or developers will start ignoring test failures or even worse disable failing tests.
Kind regards,
Markus
Regards,
Miklos
_______________________________________________ LibreOffice mailing list LibreOffice@xxxxxxxxxxxxxxxxxxxxx https://lists.freedesktop.org/mailman/listinfo/libreoffice
