On Fri, Sep 21, 2018 at 6:17 AM Chris Murphy <lists@colorremedies.com> wrote:
On Thu, Sep 20, 2018 at 8:36 PM Harold Dost <harolddost@gmail.com> wrote:
>
> Is having this as a test criteria *that* burdensome? Similar to another blocker that was being proposed, maybe it's not something that _must_ be tested, but if it's known to be broken that it should block a final. Maybe I'm missing something, but unless the image is over the capacity of a DVD, what would prevent the success of an installation?

We did have a bug a couple years ago or so, where the ISO written to
USB booted fine, but when burned to optical media wouldn't boot
anything including VM. It was a bootloader bug, if I recall correctly
- pretty sure it hit BIOS firmware only, not UEFI.

But yeah it does sound reasonable to have the same "if it's known to
be broken" then block, similar to the request for printing.

There is one difference. Those criteria that we block on yet don't test (at all or in full extent) in the QA team are usually those that are too impractical to test centrally. Ensuring that all important printer drivers work falls into that category, similarly to ensuring all important graphics cards work, etc. There's no way without community involvement we could ever cover everything important.

Optical booting is a different story, that can be tested, and quite easily (making sure it works on one system usually means it works everywhere - although there have been exceptions to the rule in the past), it just takes annoyingly long. And unlike broken printers or to a certain extent broken graphical drivers, we can't fix this with an update, we'd have to create a new compose (and make another full round of testing), which we never do. So although I'd love to get rid of optical media testing, if we keep it in the criteria, I'm not comfortable with ignoring it inside the QA team and I believe the test coverage for it would still need to be considered mandatory in that case.